Health Effects of Road Pricing In San Francisco, California · 2016-07-31 · Rajiv Bhatia Project Manager and Co‐Principal Investigator: Megan Wier Project Research Staff: Jennifer

Health Effects of Road Pricing In San Francisco, California

San Francisco Department of Public Health Program on Health, Equity and Sustainability

Technical Report: September 2011

Health Effects of Road Pricing in San Francisco, California: Findings from a Health Impact Assessment

San Francisco Department of Public Health Program on Health, Equity and Sustainability

September 2011

This work was conducted with financial support from the

Robert Wood Johnson Foundation’s Active Living Research Program.

This page left blank intentionally

Suggested Citation:

Wier M, Bhatia R, McLaughlin J, Morris D, Comerford Scully C, Harris M, Bedoya J, Cowles S, Rivard T.

Health Effects of Road Pricing in San Francisco, California: Findings from a Health Impact Assessment.

San Francisco, CA: San Francisco Department of Public Health, September 2011.

Acknowledgements:

Project Lead:

San Francisco Department of Public Health

Project Director and Co‐Principal Investigator:

Rajiv Bhatia

Project Manager and Co‐Principal Investigator:

Megan Wier

Project Research Staff: Jennifer McLaughlin,

Devan Morris, Cyndy Comerford Scully, Michael

Harris, Julian Bedoya, Stephanie Cowles, Tom

Rivard

In Coordination with:

The San Francisco County Transportation

Authority: Mobility, Access and Pricing

Study Team

Tilly Chang, Zabe Bent, Elizabeth Sall, Jesse

Koehler, Liz Brisson, Michael Schwartz

Project Consultants:

Fehr & Peers Transportation Consultants

Meghan Mitman

San Francisco Injury Center

Rochelle Dicker, Dahianna Lopez, Dharma

Sunjaya

UC Berkeley ‐ School of Public Health

Edmund Seto

Walk San Francisco

Elizabeth Stampe

We would also like to thank June Weintraub of

the San Francisco Department of Public Health

and Jon Levy of the Harvard School of Public

Health for their review and comment on this

report.

The views expressed herein do not necessarily reflect the official policies of the City and County of San Francisco; nor does mention of the San Francisco Department of Public Health imply its endorsement.


Table of Contents

Executive Summary...................................................................................................................................‐ 1 ‐

I. Background ...................................................................................................................................‐ 5 ‐

A. Health Impacts of Transportation Systems.................................................................................................. ‐ 5 ‐

B. Road Pricing: Definition and United States Policy Context .......................................................................... ‐ 6 ‐

C. Road Pricing Proposals and Studies in San Francisco, California................................................................. ‐ 7 ‐

II. Screening and Primary Objectives .............................................................................................. ‐ 11 ‐

III. Scoping........................................................................................................................................‐ 13 ‐

A. Decision Alternatives, Temporal and Geographic Boundaries................................................................... ‐ 13 ‐

B. Potential Pathways from Congestion Pricing to Health Impacts............................................................... ‐ 14 ‐

C. Scope of Health Effects & Research Questions .......................................................................................... ‐ 16 ‐

D. Data Sources.............................................................................................................................................. ‐ 18 ‐

E. Methods..................................................................................................................................................... ‐ 19 ‐

F. Health Effects Characterization and Uncertainty Assessment .................................................................. ‐ 20 ‐

IV. Baseline and Future Conditions: Populations and Traffic...........................................................‐ 21 ‐

A. 2005 → 2015: Impacted Populations........................................................................................................ ‐ 21 ‐

B. 2005 → 2015: Traffic Impacts ................................................................................................................... ‐ 24 ‐

V. Impact Analysis ...........................................................................................................................‐ 27 ‐

A. Vehicle Air Pollutants................................................................................................................................. ‐ 28 ‐

B. Road Traffic Noise...................................................................................................................................... ‐ 36 ‐

C. Active Transportation ................................................................................................................................ ‐ 43 ‐

D. Vehicle‐Pedestrian Injury Collisions ........................................................................................................... ‐ 50 ‐

E. Vehicle‐Cyclist Injury Collisions .................................................................................................................. ‐ 61 ‐

F. Health Inequities........................................................................................................................................ ‐ 66 ‐

G. Costs .......................................................................................................................................................... ‐ 70 ‐

VI. Recommendations ......................................................................................................................‐ 75 ‐

References ..............................................................................................................................................‐ 79 ‐

Appendices

Appendix A. Methods: Population Exposure Assignment for Air Quality and Noise Impacts Estimation

Appendix B. Methods: Air Pollution Impacts on Premature Mortality

Appendix C. Methods: Noise Impacts on Community Annoyance and Myocardial Infarction

Appendix D. Methods: Active Transportation and Lives Saved from Walking and Cycling

Appendix E. Methods: Vehicle‐Pedestrian Injury Collision Impacts

Appendix F. Methods and Analyses: Pedestrian Environmental Quality Index

Appendix G. Methods: Vehicle‐Cyclist Injury Collision Impacts

Appendix H. Methods: Health Effects Characterization and Uncertainty Assessment

Appendix I. San Francisco Planning Neighborhood Boundaries and Names

Attachments

Attachment A. San Francisco County Transportation Authority Mobility, Access and Pricing Study Fact Sheet,

December 2010

Attachment B. San Francisco County Transportation Authority: The San Francisco Model… In 15 Minutes,

Winter 2011


List of Tables

Table 1. Stages of Health Impact Assessment

Table 2. Key Input Data for the Road Pricing Health Impact Assessment

Table 3. Population Characteristics in 2005 (Existing Conditions) and 2015 (Projected): San Francisco,

California

Table 4. Traffic Volume (Aggregated, Segment Level) in 2005, 2015 BAU, and 2015 RP Conditions

Citywide, Outside and Inside the Potential Road Pricing Zone: San Francisco, CA

Table 5. Air Pollutants with Important Motor Vehicle Sources and Associated Health Effects

Table 6a. Estimated Traffic Attributable Fine Particulate Matter (PM 2.5) Levels and 2015 Residential

Population Exposures in 2005‐2015 BAU: San Francisco, California (n=824,000)

Table 6b. Estimated Traffic Attributable Fine Particulate Matter (PM 2.5) Levels and 2015 Residential

Population Exposures in 2015‐2015 RP: San Francisco, California (n=824,000)

Table 7. Mortality Attributable to Traffic‐related PM 2.5 in 2005, 2015 Business As Usual, and 2015 with

Road Pricing Conditions: San Francisco, California

Table 8. Road Pricing Policy Health Effects Characterization: Traffic Attributable PM 2.5 and Premature

Mortality

Table 9. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects for Traffic‐related PM

2.5 and Premature Mortality

Table 10. World Health Organization Guideline Values for Community Noise in Specific Environments

(1999)

Table 11a. Estimated Traffic Attributable Noise (Ldn) Levels and 2015 Residential Population Exposures

in 2005‐2015 BAU: San Francisco, California (n=824,000)

Table 11b. Estimated Traffic Attributable Noise (Ldn) Levels and 2015 Residential Population Exposures

in 2015 BAU‐2015 RP: San Francisco, California (n=824,000)

Table 12. Noise‐related Health Effects in 2005, 2015 Business As Usual, and 2015 with Road Pricing

Conditions: San Francisco, California

Table 13. Road Pricing Policy Health Effects Characterization: Traffic‐related Noise Health Effects

Table 14. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects of Traffic‐related

Noise

Table 15. Walk Trips and Walk Minutes Per Capita in 2005, Percent Change from 2005 to 2015 “Business

As Usual” (BAU), and Percent Change from 2015 BAU to 2015 with Road Pricing (RP): San Francisco,

California

Table 16. Bike Trips and Bike Minutes Per Capita in 2005, Percent Change from 2005 to 2015 “Business


California

Table 17. Lives Saved from Active Transportation in 2005, 2015 Business As Usual, and 2015 with Road

Pricing Conditions: San Francisco, California

Table 18. Road Pricing Policy Health Effects Characterization: Lives Saved from Active Transportation

via Walking and Cycling

Table 19. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects for Lives Saved from

Active Transportation via Walking and Cycling

Table 20. Annual Estimated Pedestrian Injury Collisions in 2005, 2015 Business As Usual, and 2015 with


Table 21. Road Pricing Policy Health Effects Characterization: Vehicle‐Pedestrian Injury Collisions

Table 22. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects for Vehicle‐

Pedestrian Injury Collisions

Table 23. Annual Estimated Vehicle‐Cyclist Injury Collisions in 2005, 2015 Business As Usual, and 2015

with Road Pricing Conditions: San Francisco, California

Table 24. Road Pricing Policy Health Effects Characterization: Vehicle‐Cyclist Injury Collisions

Table 25. Magnitude of the Estimated Health Effects for Vehicle‐Cyclist Injury Collisions: Uncertainty

Factors

Table 26. The Distribution of Traffic Density in the General Population: 2005, 2015 Business As Usual,

2015 with Road Pricing

Table 27. Traffic Density by Age: 2005, 2015 Business As Usual, 2015 with Road Pricing

Table 28. Traffic Density by Income: 2005, 2015 Business As Usual, 2015 with Road Pricing

Table 29. Economic Valuation of Health Effects: Input and Methods

Table 30. U.S. Department of Transportation Guideline Values for Lethal and Non‐Lethal Injuries

Table 31. Annual Economic Values of Transportation‐Attributable Health Effects under 2005, 2015 BAU,

and 2015 RP scenarios: San Francisco, California

List of Figures

Figure 1. The San Francisco County Transportation Authority’s Mobility, Access and Pricing Study (MAPS)

Boundaries for the “Northeast Cordon” Congestion Pricing Study Scenario: San Francisco, California

Figure 2. San Francisco County Transportation Authority Mobility Access and Pricing Study Timeline

Figure 3. Scenarios and Time Periods for HIA Analyses

Figure 4. Pathways through which Road Pricing Policies May Impact on Health

Figure 5. Average Household Income (2005, TAZ Level): San Francisco, California

Figure 6. Youth and Senior Population Densities in San Francisco, California: 2005 Conditions

Figure 7. Residential Population in 2005 and Projected Percent Change in 2015: San Francisco,

California

Figure 8. Employee Population in 2005 and Projected Percent Change in 2015: San Francisco, California

Figure 9. Traffic Volume Density in 2005, 2015 BAU, and 2015 RP: San Francisco, California

Figure 10. Health Effects of Road Pricing Policy Mediated by Air Pollution

Figure 11. Traffic attributable fine particulate matter (PM 2.5) modeled in 2005, 2015 “Business As

Usual” (BAU) and 2015 with Road Pricing (RP) Conditions: San Francisco, California

Figure 12. Pathway from Road Pricing Policy to Traffic‐Related Noise Health Impacts

Figure 13. Traffic attributable modeled noise levels in 2005, 2015 BAU, and 2015 RP: Residential Parcels

in San Francisco, California

Figure 14. Pathway from Road Pricing Policy to Physical Activity via Active Transportation

Figure 15. Activity Zones used for Active Transportation Analyses and Proportion of Residents in Each

Zone by Age: San Francisco, California

Figure 16. Pathway from Road Pricing Policy to Vehicle‐Pedestrian Injury

Figure 17a. Vehicle‐Pedestrian Injury Collisions in 2005 and 2015 “Business As Usual” (BAU) Conditions:

San Francisco, California

Figure 17b. Estimated Change in Vehicle‐Pedestrian Injury Collisions from 2005 → 2015 “Business As

Usual” (BAU) Conditions: San Francisco, California

Figure 18a. Vehicle‐Pedestrian Injury Collisions in 2015 “Business As Usual” (BAU) and 2015 with Road

Pricing (RP) Conditions: San Francisco, California

Figure 18b. Estimated Changes in Vehicle‐Pedestrian Injury Collisions in 2015 “Business As Usual” (BAU)

→ 2015 with Road Pricing (RP) Conditions: San Francisco, California

Figure 19. Annual Average Pedestrian Injury Collisions by Time of Day (2004‐2008): San Francisco,

California

Figure 20. High‐Injury Density Corridors for Vehicle‐Pedestrian Injuries (2005‐2009): San Francisco,

California

Figure 21. Pedestrian Environmental Quality in the Congestion Pricing Zone

Figure 22. Pathway from Road Pricing Policy to Vehicle‐Cyclist Injury

Health Effects of Road Pricing in San Francisco, California San Francisco Public Health Department

Program on Health, Equity and Sustainability

Technical Report: September 2011 ‐ 1 ‐

Executive Summary Substantial evidence informs us that investments and choices in the transportation sector have profound effects

on human health including on the level of physical activity, on morbidity and mortality from chronic diseases like

asthma and heart disease, and on unintentional injuries and fatalities. Furthermore, the health consequences of

transportation system infrastructure and operations are associated with significant economic costs associated with

health care and productivity. According to the recent World Health Organization policy brief "Health in the Green

Economy," infrastructure investments that shift transportation mode to walking, cycling and public transport

provide health benefits that are more holistic and of a greater magnitude than improvements in vehicle

technologies.1 These investments can also advance social and health equity for vulnerable groups, such as

children, the elderly, women, and lower‐income households. Integrating measures of health performance in

transportation planning and design processes and health impact assessments are two approaches to leverage the

activities and expenditures in the transportation sector to support population health.

This report documents the process and findings of a Health Impact Assessment (HIA) conducted by the San

Francisco Department of Public Health on one future road pricing scenario being studied by the San Francisco

County Transportation Authority (SFCTA). The scenario would charge $3 during AM/PM rush hours to travel into

or out of the northeast quadrant of San Francisco which includes a concentration of San Francisco’s currently

congested downtown streets. The scenario performed best (i.e., greatest benefits with fewest impacts) among

dozens of scenarios studied, weighing SFCTA study criteria such as change in traffic and transit congestion,

environmental benefits, and economic benefits. To date, no planning study for a specific road pricing proposal in

the United States has included a comprehensive assessment of health impacts.

The HIA team developed a causal diagram to illustrate the potential ways in which road pricing can have an impact

on health. Based on the pathway diagram, evidence of causal effects, available data, standard methods for impact

analysis and concerns expressed by local and regional stakeholders, the HIA characterizes effects related to road

pricing changes in the following health determinants: active transportation via walking and cycling; vehicle air

pollutants; road traffic noise; vehicle‐pedestrian injury; vehicle‐cyclist injury; associated economic value; and

equity implications. The HIA utilizes data on transportation and land use factors, socio‐demographic

characteristics, and health outcomes and behaviors from a variety of sources, including key inputs from the

SFCTA’s travel model. The analysis describes baseline conditions and characterizes the likelihood, severity,

magnitude and distribution of selected health effects. Where sufficient high‐quality information and methods

existed, the analysis quantifies the magnitude of health effects. The following table summarizes the key

quantitative findings from the HIA, which analyzed impacts under three scenarios: 2005 (existing conditions), 2015

BAU (business as usual – no road pricing), and 2015 RP (with road pricing).

San Francisco Public Health Department Health Impacts of Road Pricing in San Francisco, California


‐ 2 ‐ Technical Report: September 2011

Road Pricing Health Impact Assessment Findings Summary: San Francisco, California

Health Impacts

(Annual Estimates)

2005 2015 BAU 2015 RP Change:

2005 →

2015 BAU

Change:

2005 →

2015 RP

Change: 2015

BAU → 2015

RP

Overall Confidence

in Assessment

Adverse Impacts

Air Pollution, Mortality Attributable to Traffic‐related PM 2.5 (N, Ages 30 and up) High ‐ Moderate Citywide 65 66 63 1 ‐2 ‐3

Congestion Pricing Zone 24 26 23 2 ‐1 ‐3

Noise, High Annoyance (N, Ages 25 and up) HighCitywide 92,500 100,400 100,300 7,900 7,800 ‐100

Congestion Pricing Zone 36,800 40,600 40,500 3,800 3,700 ‐100

Noise, Myocardial Infarction Associated with Traffic Noise (N, Ages 30 and up) ModerateCitywide 31 34 34 3 3 0

Congestion Pricing Zone 18 20 20 2 2 0

Vehicle‐Pedestrian Injury Collisions (N, Total) High ‐ ModerateCitywide 810 860 815 50 5 ‐45

Congestion Pricing Zone 360 395 360 35 0 ‐35

Vehicle‐Cyclist Injury Collisions (N, Total) Moderate ‐ LowCitywide 270 295 290 25 20 ‐5

Congestion Pricing Zone 135 155 150 20 15 ‐5

Beneficial Impacts

Cycling for Active Transportation, Lives Saved (N, Ages 25‐64) ModerateCitywide 23 25 26 2 3 1


Walking for Active Transportation, Lives Saved (N, Ages 25‐64) ModerateCitywide 130 138 141 8 11 3


The 2005 findings summarize several of the current estimated health burdens and benefits of transportation system operation in San Francisco. Specifically:

Transportation system operations are responsible for substantial health burdens in San Francisco including impacts on life‐expectancy, heart disease, and injuries.

The adverse health consequences of transportation disproportionately burden residents within the congestion pricing zone under study.

Active transport in the city is responsible for substantial health benefits. The differences between 2005 and 2015 Business as Usual (BAU) reflect the estimated changes to health effects attributable to the transportation system with current city plans for growth and without new policies or funding to manage transportation as the population increases, which include:

worsening of traffic‐related health impacts, exacerbated in the congestion pricing zone:

• increases in air pollution‐related health impacts, including premature mortality;

• increases in traffic noise‐related health impacts, including community annoyance and heart attacks;

• increases in vehicle‐pedestrian and vehicle‐cyclist injury collisions, particularly in the congestion

pricing area; and

health benefits from increases in residents’ active transportation and associated lives saved, concentrated in

the congestion pricing zone.

The modest differences between 2015 BAU and 2015 RP reflect the estimated health impacts of the Northeast

Cordon Road Pricing Scenario, specifically:

modest health benefits from the moderation of increases in air pollution‐related mortality to below 2005

conditions;

modest health benefits from the moderation of annoyance related to noise in the congestion pricing zone

relative to 2015 BAU;




significant health benefits from increases in active transportation and lives saved from cycling and walking in

excess of benefits estimated under 2015 BAU and 2005 conditions ;

significant health benefits from the moderation of increases in vehicle‐pedestrian injury collisions under 2015

BAU to 2005 conditions, particularly in the congestion pricing zone, as well as more modest moderation of

increases in vehicle‐cycling injury collisions estimated under 2015 BAU.

The table provides a qualitative summary of our confidence in the health impacts estimates considering

uncertainties in the assumptions in the data used characterize health effects; the basis of these confidence

assessments are detailed in the report.

The analysis describes the likelihood, severity and distribution of the selected health effects, and provides an

estimate of the magnitude of these health effects attributable to transportation and the intervention of road

pricing. Analysis of the distribution of health effects is critical for the assessment of social and environmental

justice. This HIA used traffic density as a proxy for multiple adverse health effects associated with transportation

system operation, finding no inequitable effects to low‐income, elderly, or young populations defined spatially.

The HIA further estimates the economic value of the traffic‐attributable health effects (adverse and beneficial)

under the three scenarios. Total aggregate annual value of adverse health effects are substantial in all scenarios

and lowest in 2015 RP (2015 RP: $1.120 billion; 2015 BAU: $1.173 billion 2005: 2005: $1.124 billion). Injuries to

pedestrians and cyclists resulting from collisions with motor vehicles are the greatest contributor to adverse

transportation‐attributable economic costs under all scenarios – accounting for just over half of the estimated

economic costs, with premature mortality attributable to air pollution accounting for an estimated 45% and

approximately 2% associated with heart attacks attributable to traffic‐related noise. Economic value of health

benefits associated with active transportation are substantial – estimated to be somewhat greater than the

estimated adverse transportation‐related health costs in all scenarios and greatest under 2015 RP, with an

estimated $1.337 billion in 2015 RP, $1.305 billion in 2015 BAU, and $1.225 billion in 2005. As this HIA quantifies

only a limited number of health effects, the economic valuation estimates do not represent a full accounting of the

costs and benefits of transport system operation under the three scenarios.

Overall, we conclude that transportation system operation in San Francisco has highly significant health burdens

and benefits today. Health burdens are expected to increase owing to increasing motor vehicles on local roadways

in the future and increasing population densities in already congested areas. Road pricing, if implemented, could

moderate but not entirely eliminate the changes associated with a future under “business as usual” that includes

increasing populations and traffic and no new policies or funding to manage the transportation system.

Recommendations to enhance the potential health benefits of road pricing to support reductions in

transportation‐associated health costs and increases in active transportation include: a

increasing congestion pricing fees in circumstances likely to result in reduced health risks, for example, on “spare the air” days or applying specifically to more polluting vehicles;

investing in walking and biking safety improvements in locations where injuries are greatest, for example with traffic calming along arterials in and near the road‐pricing zone;

using quieter, low‐emission hybrid buses in areas where noise and air pollution are worse;

investing in walking and biking infrastructure to encourage trips by foot and by bike into and out of the road‐pricing zone;

a Notably, a number of these recommendations are under consideration by the SFCTA as programmatic improvements or mitigations to enhance travel options or mitigate/minimize traffic and related environmental impacts as part of a road pricing comprehensive program and investment package, and were not included in the travel model outputs that informed this HIA.




monitoring road‐pricing implementation to address any unanticipated traffic increases and health impacts;

encouraging active transportation and discouraging driving through more policies such as demand‐based parking fees, “unbundling” parking in new development, and transportation demand management programs.




I. Background

A. Health Impacts of Transportation Systems

Several comprehensive reviews summarize the empirical evidence linking transportation to health effects.2 Based

on these reviews, the nexus between transportation and health can broadly be organized under four domains –

access to human needs, neighborhood livability, safety, and environmental quality:

1. Access to human needs: providing (or inhibiting) access to means of livelihood (e.g., jobs), essential goods

(e.g., food, fuel and water), and essential services (e.g. health care and education);

2. Neighborhood livability: facilitating movement, physical activity and social engagement and limiting crime

and social disorder in one’s immediate neighborhood surroundings;

3. Safety: preventing injuries and fatalities in the transport system;

4. Environmental quality: preventing emissions of environmental pollution (noise, air, water) related to

system operation and associated health impacts.

Here we provide a synopsis of the health effects of transportation systems under each of these four domains. We

further elaborate on the evidence linking transportation to specific health effects analyzed in this HIA in the

Impacts Analysis section of this report (Section V).

Access to Human Needs: Transportation systems facilitate access to human needs, providing (or inhibiting) access

to means of livelihood (e.g. jobs) essential goods (e.g., food, fuel and water), and essential services (e.g. health

care and education). Transportation surveys confirm that lower income households, who are less likely to own or

have access to automobiles, take fewer travel trips than middle and higher income households.3 Disparities in

transportation resources for the poor contribute to differential access to jobs, schools, shopping, and recreation –

which varies by geographic area and factors including public transit access, service, and cost, and proximity to jobs,

goods and services.

Neighborhood Livability: The quality and design of the built environment further affect neighborhood livability,

facilitating movement, physical activity, and social engagement and potentially limiting crime and social disorder in

one’s immediate neighborhood surroundings. Neighborhoods that include pedestrian spaces or that have lower

levels of street traffic where one can walk comfortably and safely benefit health by enabling physical activity,

leisure, and social interaction.4

Safety: The surface transportation system produces a substantial share of unintentional injuries and fatalities,

particularly among more vulnerable users such as non‐motorized users. In the U.S., pedestrians and cyclists

experience a disproportionately higher risk of traffic fatality compared to vehicle occupants on both a per‐trip and

per‐mile basis.5 High volume roadways, which are associated with both higher traffic flow and higher vehicle

speeds, are primary determinants of injury frequency.6

Environmental Quality: Transport systems are a significant source of environmental pollutants such as increased

noise or emissions to water and the air. Residents living near busy roadways experience a unique combination of

environmental hazards. Grade separated highways and other high volume roadways concentrate the flow of

vehicles and are accompanied by increases in vehicle pollution emissions such as particulates, nitrogen oxides,

carbon monoxide, and benzene.7 Epidemiological research supports consistent statistical associations among

traffic proximity and several adverse respiratory health outcomes, including impairment of lung function and




asthma incidence and symptoms in children; these associations remain significant after adjustment for economic

position.8

Road traffic is the dominant source of community noise in urban areas; proximity to roadway facilities along with

vehicle type, speed and road conditions predict urban ambient noise levels.9 Chronic exposure to road traffic

noise is associated with several adverse health outcomes, including interference with thoughts and feelings,

deficits in cognitive functioning, lowered school performance, sleep disturbance, and ischemic heart disease.10

Contaminants from the development and operation of roadways, including metals and hydrocarbons, may flow

into waterways and aquifers along with water. Runoff from roadways can overwhelm the capacity of creeks or

manmade storm water drainage systems and disproportionately affect residents adjacent to roadways or those

downstream from natural and manmade drainage systems.

Disparate Adverse Impacts: Studies conducted in California show that ethnic minority and lower income

households are disproportionately represented among populations living in close proximity to busy roadways.11

This disproportionate proximity to busy roadways translates directly to disproportionate hazards and related

health outcomes. In some places, increased pedestrian collision frequencies in low‐income neighborhoods can be

explained in part by differences in traffic volume and other factors in the transportation environment.12

B. Road Pricing: Definition and United States Policy Context

Road pricing charges a fee for the use of a road facility. The fee is set based on user demand as a function based

on time of day, location, type of vehicle, number of occupants, or other factors. Road pricing is also referred to as

congestion pricing, value pricing, variable pricing, peak‐period pricing, or market‐based pricing. Road pricing that

varies with time or demand is distinct from tolling—typically a per‐use (flat) fee on motorists for a given highway

facility—which has a sole purpose of generating revenue for construction and maintenance of a facility. Road

pricing is used to account for and manage vehicle demand on the roadways and to generate revenue – while

simultaneously also achieving other goals such as reduced congestion, reduced environmental impacts (e.g.,

greenhouse gas emissions), or addressing other external costs occasioned by road users. Examples of road pricing

include HOT (High Occupancy Toll) Lanes, variable or extended parking metering, variable bridge tolls, and

congestion pricing. There are growing examples of such road and parking pricing policies nationally, and debate

over whether or not these types of policies should be implemented and how revenue should subsequently be

invested is ongoing.13 The revenue from road pricing can be reinvested in the transportation system through

capacity expansion, road maintenance and improvements, repayment for long‐term debt, public transit service, or

bicycle and pedestrian projects.14

In the United States, funds for building and maintaining transportation systems are limited and need to be used

effectively and efficiently; public investments in the transportation system can be leveraged to achieve other social

objectives. Transportation planners and policymakers have an opportunity to prevent adverse health externalities

of planning and policy decisions (reducing injuries and pollutants) and to align investments to support strategies

that promote public health (increasing accessibility and physical activity). However, such multi‐objective decision

making requires comprehensive analyses of health impacts of transportation proposals. These analyses have thus

far been largely absent from a growing number of policy debates and transportation impacts analyses in the

United States, despite potential health benefits from decreased air pollutants, traffic‐related noise levels, and

traffic collisions, and increases in active transportation. Investing road pricing revenues in transit, bike, or

pedestrian improvements could also improve accessibility to basic needs for low‐income households without

automobiles. These types of improvements are standard practice in countries where area congestion pricing is

implemented.15




Road pricing policy approaches might also have adverse health impacts. For example, increased traffic speeds

associated with reduced congestion may increase the severity of traffic injuries and pedestrians may perceive a

less safe environment for walking. Because of the multiple and potentially cumulative health impacts of traffic and

transportation planning decisions, it is vital to assess the health impacts of transportation policy decisions. A

growing number of health impact assessments in the U.S. are focused on transportation‐related planning and

policy decisions.16

Road pricing that applies to individual road facilities (HOT Lanes) has been implemented throughout the U.S. in

California, Minnesota, Colorado, Texas, and elsewhere.17 Although area‐based road pricing has not been

implemented in the United States, in 2007‐2008, area‐wide congestion pricing was considered but not

implemented in New York City. The New York City Council passed the measure and polls showed that New York

City residents backed the proposal, which provided that revenue would be invested in expanded transit service.

However, the New York State Assembly did not act on the proposal and the Legislature did not adopt authorizing

legislation in time to meet an April 2008 federal funding deadline.18 A paper analyzing this experience and

implications for road pricing in the U.S is available online.19 There have been recent reports that congestion

pricing will be reconsidered in New York City.20

Area‐based road pricing is implemented elsewhere in the world, including in London’s central business district,

Stockholm, and Singapore. Evaluation findings of area‐based congestion pricing schemes in these settings provide

evidence of potential benefits of decreased driving associated with decreased vehicle emissions, reductions in

traffic‐related noise levels, reductions in traffic collisions including those involving pedestrians, and increases in

active transportation including walking and biking.21

C. Road Pricing Proposals and Studies in San Francisco, California

The San Francisco County Transportation Authority (SFCTA) began a feasibility study of congestion pricing in San

Francisco in 2007, funded by a $1 million grant from the Federal Highway Administration. Under all congestion

pricing scenarios, motorists could choose to pay a fee to drive in specific, congested areas or corridors during peak

periods, drive at different times of day, or choose other modes of travel such as transit or biking. Revenues

generated would in turn fund transportation improvements, such as more frequent transit service, roadway

improvements, bicycle facilities and pedestrian amenities. In December 2010 the Transportation Authority Board

unanimously approved a feasibility study22 which concluded that congestion pricing could be an effective way to

manage San Francisco’s transportation system and simultaneously support the city’s future growth plans. The

Transportation Authority Board also voted to pursue additional evaluation of the concept through environmental

clearance.

In the analysis, the best performing scenarios were a Northeast Cordon (AM/PM) that would charge motorists $3

during AM/PM peak periods, and a PM‐only charge of $6 for outbound traffic in the same area.23 These scenarios

were among dozens of scenarios that were examined, each with a different mix of metrics related to

transportation system, environmental, economic, and equity impacts. The study assumed substantial up‐front

transit improvements including regional transit access improvements, e.g., BART and Caltrain, and more frequent

regional and local (Muni) express buses. The study also assumed all net revenues generated would be dedicated

to multi‐modal improvements for motorists, pedestrians, cyclists and transit riders to the cordon area.




The focus on the north eastern part of the city emerges from

the density of congestion today (average speeds on over half

of streets in downtown San Francisco areas operate below

10mph for motor vehicles and 8mph for transit vehicles).

However, several streets in the area are also high flow,

higher speed streets with mean speeds above the speed

limit. The cordon scenarios are also informed by areas

where planned growth is likely to exacerbate existing

congestion or create new availability of transit;

flexibility/feasibility to improve transit services and other

modes of travel to/from/within the area; and the

availability of net revenues generated to produce sufficient

funding to cover cost of improved services. The feasibility

study found that a Northeast Cordon (AM/PM) Scenario that

charged a $3 fee during AM/PM peak periods to drive into or

out of the northeast quadrant of San Francisco bounded by

Laguna, 18th, Guerrero streets and the waterfront (see map in

Figure 1) performed best (i.e., greatest benefits with fewest

impacts based on SFCTA criteria) among dozens of scenarios,

weighing established study criteria such as change in traffic

and transit congestion, environmental benefits, and economic

benefits. In all, the Northeast Cordon scenario would most

effectively manage demand in the city’s most congested

areas, deliver substantial net revenues, and present

manageable impacts. The next phase of analysis will refine

the system design and further evaluate benefits and impacts along the edge of the cordon, and refine the

investment package.

Lag

un

a

18th

Street

Northeast Cordon(AM/PM, $3)

Lag

un

a

18th

Street


Lag

un

a

18th

Street


Figure 1. The San Francisco County

Transportation Authority’s Mobility, Access

and Pricing Study (MAPS) Boundaries for

the “Northeast Cordon” Congestion Pricing

Study Scenario: San Francisco, California

Compared to “business as usual” conditions in 2015, the Northeast Cordon (AM/PM) scenario resulted in:

12% fewer peak period auto trips

21% reduction in vehicle hours of delay (VHD) in the Northeast Cordon

16% reduction in Northeast Cordon Greenhouse Gas Emissions (5% citywide)

$60‐80M annual net revenue for transportation services and amenities

20‐25% transit speed improvement

The complete feasibility study is accessible online.24 Additional background information regarding MAPS is in

Attachment A, the December 2010 MAPS Fact Sheet, and more detailed information is also available on the

project website: www.sfmobility.com.

The feasibility study also evaluated a representative set of discounts or exemptions, along with a representative

set of physical investments and mitigations and programmatic improvements to ensure comparable or improved

access to the cordon area. Examples of program discounts, exemptions and caps include: discounts for key groups

including disabled persons, low‐income travelers, residents and immediate abutters of zone; exemptions for

transit vehicles, taxis, and emergency vehicles; and a maximum daily cap of $6 to help mitigate impacts on

delivery‐oriented businesses, families w/school‐age children, and potentially other populations (which would be

further mitigated by programmatic enhancements to be outlined further in next phase of the MAPS evaluation).

Programmatic improvements and mitigations, such as traffic calming, streetscape and landscape improvements,



school ridesharing, and other similar programs, may further help to enhance travel options or minimize traffic;

related potential environmental impacts would be part of the comprehensive program and investment package,

but were not included in the model analysis. These discounts and program enhancements will be further

evaluated and refined in the next phase of study. The study timeline is detailed in the following Figure 2. A

decision on whether or not to implement congestion pricing is at least 2‐3 years away, and will be informed by the

environmental studies. 25

Figure 2. San Francisco County Transportation Authority Mobility Access and Pricing Study Timeline

D. Health Impact Assessment

Health impact assessment (HIA) is a structured decision‐support practice to characterize the anticipated health

effects, both adverse and beneficial, of societal decisions including projects, plans, programs, and policies

undertaken by government or the private sector. HIA makes recommendations for health‐attentive policy and

project design and may lead to more health‐responsive decisions.26 HIA follows a series of procedural steps

outlined in Table 1. The 2010 Minimum Elements and Practice Standards for HIA published by the North American

HIA Practice Standards Working Group describes criteria for quality practice that are employed in this

application.27

Table 1. Stages of Health Impact Assessment

Screening Determine need for and value of a HIA in the decision‐making process.

Scoping Determine which health impacts to evaluate, methods for analysis, and

workplan to complete the assessment.

Assessment Conduct a baseline conditions analysis and a qualified judgment of

magnitude and likelihood of potential health impacts.

Recommendations Make recommendations to manage the health impacts identified, including

alternatives to the decision, modifications to the proposed policy, program,

or project, or mitigation measures.

Reporting Complete a report of HIA findings and recommendations and communicate

results to stakeholders and decision makers.

Monitoring Track effects of HIA and decision outcomes as well as the effect of the

decision on health impacts and/or determinants of concern.








II. Screening and Primary Objectives

To date, no planning study for a specific road pricing proposal in the United States has included a comprehensive

assessment of health impacts. The feasibility study conducted by the SFCTA (described in Section I) contributes

substantially to local and national public discussion on congestion pricing and the dialogue regarding future

direction of transportation policies and investments.28 However, the feasibility study of congestion pricing did not

include a comprehensive assessment of potential health impacts. Working with SFCTA staff as well as the Mayor’s

Office and consulting several national transportation interest groups, SFDPH staff thus considered the value of an

independent HIA to the decision‐making process.

SFDPH staff determined the following factors supported the conduct of an HIA:

The congestion pricing study area affects a large share of San Francisco’s land area, residents, and

employees;

Evidence suggested that the proposal would have several potential health effects;

Several local and regional stakeholders have raised questions and concerns regarding the health and

equity impacts of road pricing on air pollution and traffic hazards to pedestrians and cyclists;

The health effects of road pricing may be important in the evaluation of policy trade‐offs;

There is little available research on the health effects of road pricing;

Existing data and methods allowed substantive analysis of health impacts of concern;

The SFCTA, the lead agency analyzing the policy, provided modeling data in support of the HIA;

There was sufficient time to conduct an HIA to inform policy design and revenue investment choices; and

The HIA could provide a model for HIA on other transportation policies or infrastructure investments at

the local, regional, state or national levels.

Based on the above factors, staff articulated the following objectives for the HIA on road pricing, consistent with

the stages of HIA.

HIA Phase Study Objectives

Scoping 1) Enumerate the scope of priority health issues and concerns related to congestion pricing in

San Francisco, California in consultation with local stakeholders and identify best available

methods to analyze these impacts;

Assessment 2) Analyze and document baseline health factors and related conditions in the targeted area;

3) Make evidence‐based judgments of the magnitude, direction, and certainty of potential health

impacts of the proposed congestion pricing policy, quantitatively and qualitatively, including

potential inequitable impacts;

4) Recommend policy modifications, decision alternatives, or other mitigations to address

potential adverse health impacts;

Reporting 5) Report findings and recommendations to local and regional stakeholders in a written report

which includes a succinct, accessible summary of the findings;




6) Report findings, recommendations and lessons learned to regional, state and national public

health, planning, and policy audiences so that the HIA conceptual approach and findings may

inform similar policy proposals throughout the country.



III. Scoping

In the Scoping stage SFDPH staff established geographic and temporal boundaries for the analysis, prioritized the

health impacts for analysis, identified the data, methods and tools for impacts analysis, identified roles for experts

and key informants, and developed a plan and timeline for external review and dissemination of findings and

recommendations.

A. Decision Alternatives, Temporal and Geographic Boundaries

The HIA compared road pricing under the Northeast Cordon scenario (2015 RP) against a business as usual (no

road pricing) alternative (2015 BAU) in the year 2015. The HIA compared the estimated health effects under each

alternative to each other and to baseline conditions without pricing in 2005; these years of analysis are the same

as those used by the SFCTA in their analyses. These scenarios are summarized in Figure 3. The geographic

boundaries of the HIA are those of the City and County of San Francisco.

Figure 3. Scenarios and Time Periods for HIA Analyses

Please Note: Throughout this report, we refer to 2015 future conditions with the following abbreviations:

2015 BAU = 2015 under “Business as Usual”: 2015 BAU is a future consistent with San Francisco policies and

investments in which future residential and employment growth is concentrated in the eastern and southern areas

of the city. These neighborhoods include the most transit accessible areas (e.g., via Muni and BART) – and also

contain the most congested streets. This growth includes large increases in residents as well as increases in

employees (summarized in Section IV of this report), and differs from previous growth in San Francisco in that this

new housing and residents are accompanied by a growing trend in “out commuting” from San Francisco to more

suburban job centers (e.g., Silicon Valley). This scenario does not include road pricing or other new policies or

funding strategies to manage the transportation system as populations increase.

2015 RP = 2015 with Road Pricing under the Northeast Cordon Congestion Pricing Scenario: 2015 RP is a future

in San Francisco with the residential and employment growth described under 2015 BAU, and additionally includes

road pricing charging motorists $3 during weekday AM/PM peak periods under the Northeast Cordon Congestion

Pricing Scenario detailed in the previous section. This scenario includes up‐front transit service enhancements.29

As described earlier in this report, the SFCTA had specified the Northeast Cordon Congestion Pricing Scenario as a

“best performing” policy for future study.





B. Potential Pathways from Congestion Pricing to Health Impacts

The HIA team developed a causal diagram (Figure 4) to illustrate the potential ways in which road pricing can

impact health. The pathways in the tan box on the right of the figure are the focus of the HIA. Empirical evidence

relating transportation and health and preliminary analysis of road pricing effects on transportation behaviors

conducted by the SFCTA (the direct impacts) informed the selection of environment effects and related health

effects.

The HIA team reviewed draft versions of the causal diagram scope with local and regional stakeholders including:

the SFCTA, the San Francisco Bay Area Metropolitan Transportation Commission, the San Francisco Bicycle

Coalition, the Environmental Defense Fund, the Sierra Club, Walk San Francisco, Livable City, the Western SoMa

Citizens Planning Task Force, the South Beach/Rincon/Mission Bay Neighborhood Association, TransForm, and

Urban Habitat. The scope was also informed by SFDPH staff participation beginning in 2008 in SFCTA‐led public

meetings, Board hearings, webinars, and technical advisory committee meetings in which representatives from the

community, local businesses, freight, sustainable transportation, local government and a number of other interests

participated.

As illustrated in Figure 4, health effects result from direct effects on travel behavior and indirect effects related to

other health beneficial impacts on the environment and on transportation systems. For example:

1. Increases in pricing may reduce the number of trips taken by automobile and increase the number of trips

taken with active transportation modes including walking and bicycling, and also increase active transportation

trips to and from public transit. Pricing could result in some people changing to alternative travel modes30 – and

therefore potentially increasing walking and biking for transportation.

2. Reductions in motor vehicle trips and miles travelled may have corresponding reductions in air pollution,

greenhouse gases, traffic‐related noise, and traffic collisions and their associated health impacts including lung and

cardiovascular diseases, stress, asthma, and traffic‐related injury and death.31 These improvements in

environmental conditions and reductions in hazards may further enable participation in active utilitarian and

leisure transportation.32

3. Congestion pricing may provide funds for transit and transportation infrastructure, including transit service

improvements and infrastructure for walkers and bicyclists, further enabling active transportation including to

public transit.33

4. Pricing‐related traffic congestion reductions may reduce delay for transit operators, improving transit service,

reliability and therefore use. These improvements carry the potential to increase the use of active transportation

to access transit.34

5. Pricing may affect travel costs, increasing transportation costs for some and reducing costs for others.

Causal diagrams for discrete pathways are included in the Impact Analysis section of this report (Section V.).



Figure 4. Pathways through which Road Pricing Policies May Impact on Health





C. Scope of Health Effects & Research Questions

Based on the above pathway diagram, evidence of causal effects, available data, standard methods for impact

analysis, and concerns expressed by stakeholders, the HIA characterizes effects related to road pricing changes in

the following health determinants:

Active transportation via walking and cycling

Air Pollutants

Noise

Pedestrian injury

Cyclist injury.

For each of these determinants, we evaluated health effects for San Francisco residents under 2015 BAU

compared to 2005 Existing Conditions and under a future scenario with Congestion Pricing under the North East

Cordon Road Pricing scheme (2015 RP). Where sufficient data were available, we quantified how changes in these

health determinants would change associated health outcomes including life‐expectancy, ischemic heart disease

events, and pedestrian and cyclist injuries.

Identifying decision impacts on health inequities, taking into account population vulnerabilities, is an explicit

purpose of HIA. Health inequities are defined as systematic disparities in health status or in the major social

determinants of health between groups with different social advantage/disadvantage (e.g., wealth, power,

prestige).35 Particular subgroups that may be particularly vulnerable to the health impacts of transportation

infrastructure and operations include: youth at high risk for physical inactivity and obesity; youth, seniors, and low‐

income populations sensitive to air pollutants and traffic‐related noise disturbances; youth at high risk of traffic

related injury, and seniors at high risk of traffic related death; and low‐income, transit‐dependent populations,

historically disproportionately burdened by the adverse health impacts of transportation planning decisions that

channel motor vehicles into their communities.36 This HIA focused on distributional impacts of alternative

scenarios on populations defined by income, age, and place.

The economic value of health costs and benefits is often considered useful information to decision‐makers. We

also assessed the economic value of a subset of health effects quantified through this assessment.

Finally based on our analysis, we considered what transportation investments or mitigations could reduce adverse

health impacts and support health benefits expected in either the 2015 BAU or 2015 RP scenarios.

A number of health‐related end points were considered in our scoping process but were not analyzed in the HIA.

These endpoints are listed below along with a rationale for their exclusion.

Impacts on San Francisco employees: With the exception of analysis of vehicle‐pedestrian and vehicle‐bicycle

collision injuries, this HIA focuses on the health impacts on San Francisco residents, though there are more than

half a million workers in San Francisco (over 40% of whom are not San Francisco residents),37 approximately two‐

thirds of whom work in the potential congestion pricing zone. Our focus on residents was largely informed by the

available health evidence and exposure‐response estimates to estimate potential health impacts related to

changes in air pollution, noise, and active transportation; health research regarding these effects is largely focused

on resident impacts. The HIA field would benefit from additional research on how transportation policies affect

employee health, in addition to the research on workers in transportation‐related professions (e.g., truck drivers).

Greenhouse gas (GHG) emissions: Climate change threatens to have global and catastrophic effects on health

through: increased frequency, intensity and length of heat waves, floods, droughts, windstorms and wildfire,




leading to increased mortality, illness and mental health impacts; increased exposures to ground‐level ozone and

aeroallergens, exacerbating cardiovascular and pulmonary illness; and shifts towards warmer temperatures,

leading to increased risk of food‐ and waterborne infectious diseases. 38 Changes in greenhouse gas emissions

were estimated as a part of the SFCTA MAPS analyses and we chose not to duplicate the analysis.

Access to local jobs, goods, and services including health care, child care: Access to local resources, including

employment, goods and services to meet daily needs, is fundamental to people meeting their basic needs for

survival and health. Through the scoping process, communities expressed concerns regarding the potentially

increasing costs of accessing these fundamental resources with the implementation of congestion pricing. The

SFCTA plans to include these specific transportation issues in the next phase of study – particularly with respect to

transportation of children to and from school and childcare.

Livability, social cohesion and associated mental and physical health impacts including stress: We did not

include these more distal health impacts in our analysis given our already broad scope and limitations in available

methods to estimate potential impacts.

Time and related stress and increased environmental exposures of being stuck in traffic or on transit: Due to

limits in currently available methods and more distal health impacts we did not estimate the health impacts of

changes in stress levels associated with changes in travel time. We do consider changes in time spent travelling via

walking and cycling. Time spent in traffic is also associated with elevated personal exposures (e.g., to air pollution

and noise), which is not taken into account in the modeling of environmental exposures.

Transportation costs and low‐income drivers: The SFCTA analysis considered the broad equity implications of who

would be paying the fee. While we (and SFCTA staff) acknowledge that equity is a serious concern for local and

regional stakeholders, we did not focus on these larger equity issues given the SFCTA’s current policy

recommendations to include discounts for low‐income drivers and/or transit riders, as well as additional findings

that less than 5% of peak‐period travelers to downtown San Francisco are drivers with annual household incomes

< $50,000. Furthermore, data show many low‐income travelers in the Bay Area already rely on public transit and

would thus benefit from faster, more reliable transit travel times resulting from pricing fee investments. In a 2007

poll of regional travelers, the SFCTA further found that support for studying a San Francisco congestion pricing

program is highest among low‐income Bay Area residents.

Economic Impacts on Local Businesses: This issue was commonly raised in public forums, was addressed in the

SFCTA’s feasibility study, and is further described in Section 2.4 of the feasibility study report. Economic effects on

businesses were not represented by commenters as health concerns.

Fee Investments: Based on available data regarding the policy, our analysis focuses on health impacts resulting

from changes in travel behavior. Additional impacts are anticipated based on investments of the fee though those

specifics will be determined in the next phase of the study. We therefore focused our recommendations on how

investments could be leveraged to promote health.

Regional Impacts: Our HIA focuses on local impacts within the City and County of San Francisco, though

congestion pricing proposals also impact on transportation behaviors and vehicle miles travelled at a regional level.

Notably, the SFCTA’s transportation analysis for the feasibility study was a regional analysis, and the cordon area

includes local residents as well as regional workers.




D. Data Sources

Based on the framework established in Figure 4, our HIA addresses the aforementioned research questions

regarding the health impacts of transportation policy decisions on residents in the City and County of San

Francisco, California based on data from three scenarios: 2005 Existing Conditions, 2015 BAU, and 2015 RP. An

overview of the variables used for analyses, data sources, and unit of analysis is detailed in the following table.

Table 2. Key Input Data for the Road Pricing Health Impact Assessment

Variable Data Source Geographic Unit of Analysis HIA Analyses

Transportation and Land Use Conditions: 2005 , 2015 BAU, 2015 RP

City Lots, Building Heights, Zoning

Designation

SF Planning Lot AQ, Noise

Pedestrian Environmental Quality PEQI Data Collection Street Segment, Intersection Ped Inj

Traffic Volume by Vehicle Type, Time of Day,

Traffic Speed (Free Flow and Congested

Conditions), Bus Volumes

SFCTA Model Street Segment AQ, Noise, Cyclist Inj,

Ped Inj, Equity

Trips, Travel Mode by Age Category BATS Bay Area Region Active Transport

Walk and Bike Trips, Travel Time SFCTA Model Transportation District Active Transport, Cyclist

Inj

Socio‐demographic Conditions: 2005 and 2015

Resident Age AGS Inc. Estimates Census Block All

Average Household Income ABAG Estimates Transportation Analysis Zone Equity

Resident Population Data ABAG Estimates Transportation Analysis Zone All

Health Outcomes and Behaviors: 2005 Existing Conditions

Mortality CDPH (Vital Statistics) County AQ, Active Transport

Myocardial Infarction, Hospital Admission CDPH (OSHPD) County Noise

Vehicle‐Pedestrian Injury Collisions, Vehicle

Cyclist Injury Collisions

SWITRS Intersection, Census Tract, County Ped Inj, Cyclist Inj

CDPH: California Department of Public Health

OSHPD: Office of Statewide Health Planning and Development

PEQI: Pedestrian Environmental Quality Index

SFCTA: San Francisco County Transportation Authority

ABAG: Association of Bay Area Goverments

AGS, Inc.: Applied Geographic Solutions, Incorporated

AQ: Air Quality

BATS: Bay Area Travel Survey

Cyclist Inj: Vehicle‐Cyclist Injury Collisions

Ped Inj: Vehicle‐Pedestrian Injury Collisions

The SFCTA’s travel forecasting model – SF‐CHAMP 4 ‐ estimates changes in travel patterns in San Francisco and the

broader nine‐county Bay Area under different land use, population, and transportation systems conditions (see

Attachment B for an overview of the model).39 SF‐CHAMP 4 model inputs include land use, transportation, and

population data from the SF Planning Department, Association of Bay Area Governments, U.S. Census,

Metropolitan Transportation Commission, and other sources. Model outputs include street level traffic volumes

and speeds, transportation district‐level trips by mode (walk/bike/transit/driving), travel duration, and district of

residence of trip maker (n=12 San Francisco districts). We used these SF‐CHAMP transportation model outputs for




the 2005, 2015 BAU, and 2015 RP conditions, provided by SFCTA staff, as key inputs into the HIA analyses, where

transportation parameters were necessary. An important caveat is that these inputs are also modeled outputs of

the SFCTA’s travel forecasting model. While the model is internationally regarded as a sophisticated travel

forecasting approach which provides the best available estimates, its outputs are not precise predictions.

We also used Bay Area Travel Survey regional data on trip making and travel mode by age, as data was not

available just for San Francisco, to estimate the proportion making trips and transportation mode by age.40

We accessed City Parcel Data (city lots), Building Data (height) and Zoning Data (Land use designation) from the SF

Planning Department for more detailed spatial assignment of residential populations as described in Appendix A.

We used data collected with SFDPH’s Pedestrian Environmental Quality Index (PEQI) – an observational data

collection tool developed by SFDPH to assess physical environmental features that support walking in San

Francisco ‐ to assess existing conditions in and proximate to the area of focus for the congesting pricing study. The

PEQI integrates 30 variables that reflect the quality of the built environment for pedestrians;41 PEQI methods are

further detailed in Appendix F.

We used the same data as the SFCTA for existing and future resident and employee populations and average

household income, based on regional estimates and projections provided to the City and County of San Francisco

by the Association for Bay Area Governments, the regional planning agency for the San Francisco Bay Area. We

used socio‐demographic data on age and sex from census‐block level estimates and projections purchased by the

City and County of San Francisco for existing and future conditions.42 As noted above, an important caveat is that

these are the best available future estimates and not precise predictions.

We used data on collisions resulting in injury to pedestrians and bicyclists in the City and County of San Francisco

from California’s Statewide Integrated Traffic Records System (SWITRS) which contains data on reported vehicle

collisions on public roadways.43 We used data on resident hospitalizations for myocardial infarction from the

California Department of Public Health’s Office of Statewide Health Planning and Development, and mortality data

from the California Department of Public Health’s Office of Vital Records.

Data caveats, uncertainties, and implications for health effects estimates are further discussed in the Impacts

Analysis section for each specific health impact analyzed in this report; this approach to health effects

characterization and uncertainty assessment is further explained in Appendix H.

E. Methods

We used the above data and ArcGIS 10, Stata 9, Microsoft Access, and Microsoft Excel software to conduct the

analyses for this HIA. We include detailed summaries of the HIA methods used for each of the following key HIA

outputs in the Appendices of this report. These methods are also summarized in their respective subsections in

the Impact Analysis section of this report (Section V).

Appendix A. Methods: Population Exposure Assignment for Air Quality and Noise Impacts Estimation

Appendix B. Methods: Air Pollution Impacts on Premature Mortality

Appendix C. Methods: Noise Impacts on Community Annoyance and Myocardial Infarction



Appendix F. Methods and Analyses: Pedestrian Environmental Quality Index






F. Health Effects Characterization and Uncertainty Assessment

Using guidance provided in Health Impact Assessment: A Guide for Practice, we characterized our estimated health

effects based on our judgment of whether they were plausible, based on sound evidence, and logical, and also

acknowledged data limitations and uncertainties.44 Specifically, we described characteristics of the estimated

health effects including likelihood, severity, magnitude, and distribution. The likelihood of an effect represents the

degree of certainty that it will occur; severity of a health effect indicates its importance and intensity; magnitude

represents how much a health outcome might change as a result of a decision course of action; and the

distribution of effects reflects whether they are shared fairly among the affected populations. We also assessed

each health effect characterization with respect to certainty in the baseline frequency of disease, assessment of

exposure, and the statistical relationship between exposure and disease used to make health impacts predictions.

Appendix H includes an excerpt from Health Impact Assessment: A Guide for Practice which provides a detailed

summary of this approach to health effects characterization and uncertainty assessment.



IV. Baseline and Future Conditions: Populations and Traffic

In this section we describe the existing conditions in 2005 and the predicted future 2015 conditions with respect to

local populations and traffic in San Francisco, California, key inputs for the HIA estimates of policy‐related health

impacts.

A. 2005 → 2015: Impacted Populations

Based on our framework (Figure 4), we expect that the communities most likely to experience changes to health‐

relevant environmental conditions under BAU

compared to a future in 2015 with congestion

pricing are the residents within and proximate to

the area boundaries. The congestion pricing area

depicted in red in the maps in this section includes

the northeast and downtown areas of San

Francisco – bordered by the I‐80 coming off of the

Bay Bridge with off‐ramps that channel commuter

traffic onto local streets, 18th street to the south,

and Laguna on the west. The following Table 3

compares the overall population composition in

the congestion pricing area with that outside the

pricing area based on a number of key population

demographic indicators estimated for 2005 and

2015 conditions. This potential congestion pricing

zone is comprised of a diversity of neighborhoods

including San Francisco’s downtown business

district, Union Square, North Beach, Russian Hill, Chinatown, the Marina, Pacific Heights, Nob Hill, Downtown Civic

Center, the Tenderloin, off/on ramps from the Bay Bridge, as well as South of Market, Mission Bay, and northern

Potrero Hill (a map including neighborhood boundaries and neighborhood names is included in Appendix I). These

neighborhoods are diverse in terms of socioeconomic status, representing some of the poorest and wealthiest

communities in the city – with an evident concentration of low income communities in the congestion pricing area

(Figure 5). While a lower proportion of children and youth live in the congestion pricing area (11%, 2005)

compared to the area outside the zone (18%, 2005) – there are evident concentrations of children and youth in the

area, particularly in and around Chinatown and the Tenderloin communities (Figure 6). A slightly higher

proportion of the population in the congestion pricing area is aged 65 and older (16%, 2005) compared to

populations living outside of the area (13%, 2005).

Figure 5. Average Household Income (2005, TAZ

Level): San Francisco, California





Table 3. Population Characteristics in 2005 (Existing Conditions) and 2015 (Projected): San Francisco,

California

Population Characteristics 2005 2015 2005 → 2015

(% Change )

Citywide 796,000 824,000 4%Outside Pricing Zone 624,000 641,000 3%In Pricing Zone 172,000 183,000 6%

Citywide 553,000 637,000 15%Outside Pricing Zone 188,000 208,000 11%In Pricing Zone 365,000 429,000 18%

Citywide< $35,000 14% 13% ‐1%

$35,000 ‐ 70,000 60% 57% ‐3%>$70,000 26% 30% 4%

Outside Pricing Zone< $35,000 3% 3% ‐1%

$35,000 ‐ 70,000 71% 67% ‐4%>$70,000 25% 31% 5%

In Pricing Zone< $35,000 41% 38% ‐3%

$35,000 ‐ 70,000 32% 34% 3%>$70,000 27% 28% 1%

Citywide 4% 4% 0%Outside Pricing Zone 5% 5% 0%In Pricing Zone 3% 3% 0%

Citywide 12% 11% ‐1%Outside Pricing Zone 13% 12% ‐1%In Pricing Zone 8% 7% ‐1%

Citywide 13% 15% 2%

Outside Pricing Zone 13% 14% 1%

In Pricing Zone 16% 17% 1%

Citywide 46% na naOutside Pricing Zone 46% na naIn Pricing Zone 44% na na

People of Color (% of population)

Employees (N)

Residents (N)

Average Household Income (%, TAZ level)

Children (% of population, age 0‐4)

Youth (% of population, age 5‐19)

Seniors (% of population, age 65 and older)

The area of focus for congestion pricing draws a diversity of employees and visitors; this area includes the Financial

District, City Hall and numerous local government buildings, Fisherman’s Wharf, shopping and theater districts,

and AT&T Park (home of the San Francisco Giants baseball team). As detailed in Table 3, approximately two‐thirds

of individuals employed San Francisco work in the congestion pricing target area.



Figure 6. Youth and Senior Population Densities in San Francisco, California: 2005 Conditions

As depicted in the maps in Figures 7 ‐ 8, in 2015 there are projected increases in both residential and employee

populations in San Francisco, most notably in the southeast section of the potential congestion pricing zone area

(outlined in red). Specifically, we see resident population increases from 2005 to 2015 of 4% citywide, the increase

in the congestion pricing area (6%) double that in areas outside the zone (3%). Employment increases in 2015 area

also more concentrated in the congestion pricing study zone (Figure 8). The potential congestion pricing zone

accounts for approximately 17% of the city’s land area, 22% of San Francisco’s residents, 66% of San Francisco

employees, and approximately one‐third of the city’s aggregate traffic volume in the study scenarios (see next

section).

Figure 7. Residential Population in 2005 and Projected Percent Change in 2015: San Francisco,

California





Figure 8. Employee Population in 2005 and Projected Percent Change in 2015: San Francisco,

California

B. 2005 → 2015: Traffic Impacts Based on the SFCTA model SF‐CHAMP, there are over 4 million trips to, from, and within the city each day in 2005

conditions – approximately half of which are made by people starting trips from outside of the city. The following

maps (Figure 9) depict traffic density and changes in traffic density under the three study scenarios. Map A

illustrates traffic density in 2005 conditions, while Map B depicts changes in traffic density from 2005 – 2015 BAU,

illustrating increases in traffic on the local streets in the already heavily trafficked congestion pricing target area as

well as along San Francisco’s freeways as regional traffic increases. Map C and Map E illustrate traffic density in

the 2015 scenarios under business and usual versus with road pricing; Map D visualizes that spatial distribution of

changes in traffic density between 2015 BAU and 2015 RP. As evident in Map D, relative to traffic density in 2015

BAU, there are traffic increases in 2015 RP on the southern edges of the road pricing boundary in the Castro area

of San Francisco as well as in the Mission. There is also an increase along the northern part of Highway One which

runs north and south along the western side of the City, near the Presidio and the exit off the Golden Gate Bridge.



Figure 9. Traffic Volume Density in 2005, 2015 BAU, and 2015 RP: San Francisco, California

Map A. 2005 Traffic Density Map B. Change in Traffic Density: 2005 → 2015 BAU

Map C. 2015 RP Traffic Density Map D. Change in Traffic Density: 2005 → 2015 RP

Map E. 2015 BAU Traffic Density Map F. Change in Traffic Density: 2015 BAU → 2015 RP





As detailed in the following table, from 2005 to 2015 BAU, traffic citywide is estimated to increase approximately

7% ‐ with the increase in the congestion pricing zone (10%) double that in areas outside the zone (5%). With the

introduction of road pricing in 2015, traffic is estimated to decrease citywide by 4%, with traffic reductions largely

in the pricing zone (11% vs. <1% outside the pricing zone).

Table 4. Traffic Volume (Aggregated, Segment Level) in 2005, 2015 BAU, and 2015 RP Conditions

Citywide, Outside and Inside the Potential Road Pricing Zone: San Francisco, CA Traffic Volume (Aggregated) 2005 2015 BAU 2015 RP 2005 → 2015

BAU (%

Change )

2005 → 2015 RP

(% Change )

2015 BAU →

2015 RP (%

Change )Citywide 80,032,000 85,431,000 81,800,000 7% 2% ‐4%Outside Pricing Zone 52,675,000 55,346,000 55,108,000 5% 5% ‐0.4%In Pricing Zone 27,357,000 30,085,000 26,692,000 10% ‐2% ‐11%

Notably, land use and transportation policy decisions in San Francisco have directed increased residential densities

along central city transportation corridors and on lands that were formerly industrial uses.45 Several of the areas

experiencing increases in residential density have high existing traffic densities, yet a low level of car ownership

and higher levels of local and regional transit access (e.g., the South of Market). Other locations with planned

growth are more vehicle‐oriented ‐ further from the downtown core with few local businesses and retail services

(e.g., Potrero Hill and Mission Bay). As detailed in Figure 7 and Table 4, most of this new residential development –

and associated increases in vehicle trips under BAU conditions ‐ are more concentrated in the congestion pricing

zone. Forecasted increases in residents and in traffic in growth areas result in increased conflicts among

residential and transportation uses. Changes in transportation‐attributable health effects under 2015 BAU

conditions reflect these conflicts.




V. Impact Analysis

This section summarizes the results of our health impacts analysis. The subsections are organized around health

determinants selected in scoping: vehicle air pollutants, traffic‐related noise, active transportation, vehicle‐

pedestrian injury collisions, and vehicle‐cyclist injury collisions. For each determinant, we provide:

Discussion of causal effects related to exposure factors and health determinants

Analysis of changes in proximate exposure factors and health determinants

Quantitative analysis of one or more health outcomes

Overall characterization of health effects

Certainty in health effects characterization

We also include separate subsections assessing health equity with respect to traffic exposure as measured by

traffic density and costs of traffic‐related health impacts where monetary valuations are available.

As noted in the earlier discussion regarding the HIA data sources and methods, there are two caveats relevant for

all of the analyses:

o Forecasted health effects are based on estimated changes in transportation conditions and populations

(as described in Section III) estimated with the best available models, however actual population and

transportation conditions, and thus associated health effects, may ultimately vary from these estimates.

Errors in population and transportation forecasts are not likely to be different in proportion or magnitude

among areas inside vs. outside the congestion pricing zone, providing more confidence regarding the

estimated relative differences in health effects among the different scenarios.

o Health impacts of 2015 RP are estimated based on SFCTA SF‐CHAMP model outputs that do not include

programmatic improvements and mitigations that would be implemented with funding from road pricing

subsequent to programmatic implementation or because of requirements for programmatic

environmental mitigations. These improvements and mitigations might include traffic calming,

streetscape and landscape improvements, and school ridesharing. These enhancements will be further

evaluated in the next phase of the SFCTA’s study.

A final note is that we refer to the area that would be included in the potential Northeast (AM/PM) Cordon

Scenario boundaries as the Congestion Pricing Zone in the tables reporting the results of the analyses of 2005,

2015 BAU, and 2015 RP scenarios. Notably, the only scenario in which there is a theoretical road pricing zone and

associated boundaries is the 2015 RP scenario. Under 2005 and 2015 BAU conditions, we thus use the term

"Congestion Pricing Zone" (or “Road Pricing Zone,” in the text) solely as a spatial reference to the area included in

the potential Northeast Cordon Scenario boundaries.




A. Vehicle Air Pollutants

Health effects of roadway vehicle air pollutants

Engine exhaust from diesel and gasoline engines in roadway vehicles is a complex mixture of particles and gases.

Air pollutants in vehicle exhaust include carbon monoxide (CO), particulate matter (PM), and nitrogen oxides

(NOx), as well as other non‐criteria toxic air contaminants such as benzene, 1,3‐butadiene, formaldehyde,

acetaldehyde, acrolein, naphthalene, and diesel exhaust. Particulate matter, carbon monoxide, nitrogen dioxide,

and ozone have well‐established causal relationships with human health and are subject to nationwide ambient air

quality standards, monitoring and control requirements under the Federal Clean Air Act.46

This report did not focus on the effects of other vehicle emissions, including ozone precursors and fine particulate

matter which have important health effects over a broader spatial area. The location of roadways and other

transportation facilities – and the amount of traffic associated with those facilities ‐ determines the spatial

patterns of exposure to several traffic‐related air pollutant emissions from vehicle sources in urban areas. A 2007

meta‐analysis based on 33 exposure studies and four pollutants ‐ carbon dioxide, nitrogen oxides, particulates and

ultra‐fine particulates ‐ found significant spatial differences exist in multiple traffic‐related air pollutants relative to

proximity to busy roadways.47 Research in some locations based on measurements of fine particulate matter

(PM2.5) found that a significant share of spatial intra‐urban air pollution variation in ambient levels of PM2.5 is due

to local traffic sources,48 and that traffic density explains variation in local and regional PM2.5 concentrations.49

Studies on traffic proximity assess the effects of exposure to traffic‐related air pollutants collectively, as a

mixture.50 A Health Effects Institute (HEI) Report in 2008 concluded that “Evidence was ’sufficient‘ to infer a causal

relationship between exposure to traffic‐related air pollution and exacerbation of asthma and ’suggestive‘ to infer

a causal relationship with onset of childhood asthma, non‐asthma respiratory symptoms, impaired lung function,

and total and cardiovascular mortality.”51 Individual epidemiological studies associate roadway proximity or

vehicle emissions to impairments of lung function; 52 asthma symptoms; 53 medical visits for asthma;54 asthma

prevalence and incidence; 55 and ischemic heart disease.56

Figure 10 details the pathway through which road pricing policy contributes to air pollution and related health

effects. Specifically, road pricing contributes to decreases in auto trips and resulting local traffic volumes as well as

changes in local traffic speeds. This in turn results in changes to vehicle pollution emissions and an associated

range of health effects.



Figure 10. Health Effects of Road Pricing Policy Mediated by Air Pollution

The pathway in Figure 10 is specific for PM2.5 and premature mortality but the framework also applies to other

vehicle emissions and health effects. Pollutants with important vehicle sources are identified in Table 5. Reviews

of causal evidence linking individual pollutants and health outcomes can be found in several authoritative US

Environmental Protection Agency (USEPA) scientific reviews including those conducted by the Clean Air Scientific

Advisory Committee (CASAC) on Clean Air Act criteria pollutants.57





Table 5. Air Pollutants with Important Motor Vehicle Sources and Associated Health Effects

Air Pollutant Source Health Effects

Ozone Tropospheric ozone is formed in the atmosphere from chemical transformation of certain air pollutants in the presence of sunlight. Ozone precursors include vehicles, other combustion processes and the evaporation of solvents, paints, and fuels

Eye irritation, airway constriction, and shortness of breath and can aggravate existing respiratory diseases such as asthma, bronchitis, and emphysema.

Carbon Monoxide (CO)

Produced due to the incomplete combustion of fuels, particularly by motor vehicles

Reduced oxygen‐carrying capacity of the blood resulting in fatigue, impaired central nervous system function, and induced angina.

Particulate Matter

(PM10 and PM2.5)

Diverse sources including motor vehicles (tailpipe emissions as well as brake pad and tire wear, wood burning fireplaces and stoves, industrial facilities, and ground‐disturbing activities

Impaired lung function, exacerbation of acute and chronic respiratory disease, including bronchitis and asthma, emergency room visits and hospital admissions, and premature death.

Nitrogen Dioxide (NO2)

Combustion processes in vehicles and industrial operations

Acute and chronic respiratory disease

Criteria Pollu

tants

Sulfur Dioxide (SO2)

Combustion of sulfur‐containing fuels such as oil, coal, and diesel

Acute and chronic respiratory

Diesel exhaust Diesel engines Probable human carcinogen (IARC Group 2A)

Benzene Gasoline engines Known human carcinogen (IARC Group

1A)

1,3 butadiene Motor vehicle engines Probable human carcinogen (IARC Group

2A)

Non‐criteria Pollu

tants

Benzo(a) pyrene

Motor vehicle engines Probable human carcinogen (IARC Group

2A)



Analysis of Traffic‐Attributable Fine Particulate Matter (PM2.5) and Mortality

In general, reductions in air pollution sources in populated areas result in relatively proportional reductions in air

pollution exposure and air pollution‐attributable health effects. Traffic represents only one source of air pollution,

thus the reductions in pollutant concentrations associated with a change in traffic trips and volume depends on

the traffic‐attributable share of the pollutant. For example, research shows that local traffic sources are

responsible for roughly 20‐30% of the intra‐urban variation in respirable particulate matter and roughly 60‐70% of

the intra‐urban variation of Nitrogen Dioxides in urban areas in the United States.58

While it is possible to estimate quantitatively the effects on multiple health outcomes for several distinct

pollutants, we chose to apply estimation methods only to one pollutant‐outcome relationship: premature

mortality associated with chronic exposure to PM2.5. We selected this particular outcome for the following

reasons: the severity of the health outcome (pre‐mature mortality); the established causal relationship between

exposure to PM2.5 and pre‐mature mortality; the availability of exposure‐response functions utilized in existing

regulatory impacts assessment by the USEPA and the California Air Resources Board; and existing SFDPH technical

capacity to model community‐level traffic attributable changes in PM2.5 concentrations. Detailed methods for the

quantitative assessment are provided in Appendix B. The fact that we limited our analysis to one health outcome

means that the quantitative estimates do not capture the full spectrum of health effect of changes in air

pollution in the different scenarios.

The following maps (Figure 11) depict traffic‐attributable PM2.5 in three scenarios: 2005, 2015 BAU, and 2015 RP.

We estimated increases in the highest levels of PM2.5 from 2005 (upper left) to 2015 BAU (upper right) in the area

inside the road pricing boundary and along the freeways and highways. When comparing 2015 RP (lower right) to

2015 BAU (upper right) we see minor decreases within the cordon particularly in areas with relatively higher

baseline concentrations along with minor increases in the areas just outside of the cordon boundary due to

moderate increases in traffic volumes on the edges of the zone as detailed in Figure 9, Map D.

Figure 11. Traffic attributable fine particulate matter (PM2.5) modeled in 2005, 2015 “Business As

Usual” (BAU) and 2015 with Road Pricing (RP) Conditions: San Francisco, California





Table 6a and 6b provide a tabular summary of the differences among the scenarios documenting the proportional

increase or decrease in share of the San Francisco population exposed to varying levels of traffic attributable PM2.5.

Table 6a provides estimates of this change for 2005 → 2015 BAU conditions and Table 6b provides the estimates

for 2015 BAU → 2015 RP conditions.

Table 6a. Estimated Traffic Attributable Fine Particulate Matter (PM2.5) Levels and 2015 Residential

Population Exposures in 2005‐2015 BAU: San Francisco, California (n=824,000)

2005 0.1‐0.2 0.3‐1.0 1.1‐2.0 2.1‐3.0 3.1‐4.0 4.1‐6.7

0.1‐0.2 0.2% ‐ ‐ ‐ ‐ ‐0.3‐1.0 0.0% 26% 0.5%

0.6%0.6%

0.8%

3%

‐ ‐ ‐

1.1 ‐2.0 ‐ 0.9% 36% ‐ ‐2.1‐3.0 ‐ ‐ 1.1% 23% ‐3.1‐4.0 ‐ ‐ ‐ 0.4% 7%4.1‐6.7 ‐ ‐ ‐ ‐ ‐ 4%

95%2%

No ChangeReduction from 2015 BAU with 2015 RPIncrease from 2015 BAU with 2015 RP

SUMMARY: 2015 Residents Experiencing Impacts (%)

2015 BAU

Overall, under 2015 BAU relative to 2005, 95% of the population does not see a substantial change in PM2.5

exposure. However, 2% of residents would experience decreases in exposure relative to 2005 levels, and 3%

would have increases in exposure relative to 2005 levels. Importantly, residents experiencing increases in

exposure tend to be those already in higher exposure areas in 2005.

Table 6b. Estimated Traffic Attributable Fine Particulate Matter (PM2.5) Levels and 2015 Residential

Population Exposures in 2015‐2015 RP: San Francisco, California (n=824,000)

2015 BAU 0.1‐0.2 0.3‐1.0 1.1‐2.0 2.1‐3.0 3.1‐4.0 4.1‐6.7

0.1‐0.2 0.2% 0.0%0.7%

0.7%0.4%

2%

‐ ‐ ‐ ‐0.3‐1.0 0.0% 26% ‐ ‐ ‐1.1 ‐2.0 ‐ 1.0% 36% ‐ ‐2.1‐3.0 ‐ ‐ 3% 20% ‐3.1‐4.0 ‐ ‐ ‐ 4% 4% ‐4.1‐6.7 ‐ ‐ ‐ ‐ 2% 3%

88%10%

No ChangeReduction from 2015 BAU with 2015 RPIncrease from 2015 BAU with 2015 RP

SUMMARY: 2015 Residents Experiencing Impacts (%)

2015 RP

Under 2015 RP relative to 2015 BAU exposure levels, 88% of the population does not see a substantial change in

PM2.5 exposure. Ten percent of residents would have decreases in exposure relative to 2015 BAU levels, and these

decreases are experienced across the range of 2015 BAU exposure levels including the highest exposure category

(4.1‐6.7). 2% of residents would have increases in exposure in 2015 RP relative to 2015 BAU levels.

Using the estimated changes in concentrations of PM2.5 (methods detailed in Appendix B) and the population

exposure assignment method detailed in Appendix A, we estimated the impact of pollutant concentration changes

on pre‐mature mortality. As detailed in the following summary, we estimate an increase in premature deaths

attributable to PM2.5 from 2005 – 2015 BAU, with an increase of one death citywide and two deaths in the cordon

zone (the larger increase in deaths in the congestion pricing zone relative to citywide explained by variation in



estimated net changes in PM2.5 exposure for residents across the city largely explained by changes in vehicle

volume). Under 2015 RP, we estimate a reduction of three deaths annually compared to BAU conditions – all in

the area that is the focus of the congestion pricing policy.

Table 7. Mortality Attributable to Traffic‐related PM2.5 in 2005, 2015 Business As Usual, and 2015 with


Health Impacts

(Annual Estimates)

2005 2015 BAU 2015 RP Change:

2005 →

2015 BAU

Change:

2005 →

2015 RP

Change: 2015

BAU → 2015

RP

Air Pollution, Mortality Attributable to Traffic‐related PM 2.5 (N, Ages 30 and up)Citywide 65 66 63 1 ‐2 ‐3

Congestion Pricing Zone 24 26 23 2 ‐1 ‐3

Health Effects Characterization: Air pollutant Health Effects

Table 8 summarizes our overall characterization of the health effects of the road pricing policy under study on

premature mortality predicted by traffic‐attributable PM2.5. Based on the available evidence: there is a high

likelihood that the proposed policy will reduce pre‐mature mortality due to respirable particulate matter exposure

health; the severity of the effect— including acute and chronic respiratory diseases and mortality—is high; the

magnitude of the predicted decrease in premature deaths is limited relative to 2015 BAU; and the reduction of

deaths particularly in the cordon zone contributes to a restorative equity effect.





Table 8. Road Pricing Policy Health Effects Characterization: Traffic Attributable PM2.5 and Premature

Mortality

Health Effects:

Characteristics

Interpretation Characterization

Overall Assessment Based on the following health effects characterization and

uncertainty factors regarding the magnitude of estimated

effects, what is the overall assessment of confidence in the

health effects estimate?

High‐Moderate

Likelihood How certain is it that the road pricing policy under study will

affect traffic‐related premature mortality, irrespective of the

frequency, severity, or magnitude of the effect?

Very Likely/Certain: Consistent evidence

for causality from epidemiologic studies

with diverse populations, pollution mixes,

and exposure range, as well as evaluation

studies of congestion pricing in other

countries.

Severity How important is the effect on traffic‐attributable (PM2.5)

premature deaths with regards to human function, well‐being

or longevity, considering the affected community's current

ability to manage the health effects?

High: The impacts on PM2.5 are associated

with reductions in mortality.

Magnitude How much will traffic‐attributable (PM2.5) premature deaths

change as a result of the road pricing policy under study?Limited (High‐Moderate Certainty)a:

Compared to 2015 BAU, 2015 with RP

contributes to a decrease in 3 traffic‐

attributable (PM2.5) premature deaths

each year.

Distribution Will the effects be distributed equitably across populations?

Will the road pricing policy under study contribute to a reversal

of baseline inequities?

Restorative Equity Effects: Populations in

the cordon zone have been historically,

disproportionately burdened with traffic‐

attributable air pollution in their

communities. The road pricing policy

under study has the potential to reduce

some of this inequity and related deaths.

a See Table 9 for a summary of factors impacting on the certainty of the magnitude of the health effects characterization.

Certainty in Magnitude of Health Effects

The quantitative analysis of the magnitude of the impacts on premature mortality includes a number of

assumptions and parameters with uncertainty as summarized in Table 9. These uncertainties informed the level of

confidence of the health effect characterization summarized in Table 9. Two major uncertainties relate to

predicting exposure and the validity of the exposure response function.

Exposure analysis depends on accuracy both in traffic changes as well as the emissions model. As stated in Section

III, we used SF‐CHAMP model outputs as our best available estimates of traffic in the three scenarios. The EMFAC

2007 emissions model used in this analysis accounts for vehicle types and emissions in current and future years.

EMFAC 2007 does not fully account for emissions reductions likely to occur in the future, for example, with

California regulations on greenhouse gas emissions from passenger vehicles authorized by California Assembly Bill

1493 (Pavley). Estimation of the impacts of these regulations has been largely focused on greenhouse gases as

opposed to fine particulates. A conservative assumption made for this analysis is that the proportion of traffic

comprised of trucks does not change across time or scenarios. Also, given the nature of the San Francisco bus

fleet, bus volumes are not included in the traffic inputs. Given anticipated fleet conversions to alternative fuels,59



we do not anticipate changes in on‐road transit vehicles would contribute to significant changes in exposure and

resultant health impacts.

The exposure‐response function (ERF) is based on studies conducted in diverse contexts and is applicable to the

San Francisco context. However, the studies used to develop the ERF examine the effects on health of inter‐

regional differences in pollutants. This analysis applies those ERFs to within‐region concentration differences.

Some have argued that ERFs based on inter‐regional studies may under‐estimate the slope of the concentration‐

response relationship.60 We also assume that we can extrapolate the ERF to the exposure range found in San

Francisco. The range of exposure in our analysis is within the range of the studies used to define a E‐R relationship

for PM2.5. Scientific evidence has not established a threshold below which effects of PM2.5 on mortality do not

occur. The ERF does not account for differential vulnerability. Vulnerability or resilience factors to air pollution

health effects may vary within study area. Our approach does not account for socio‐economic factors including

poverty, education, and employment that may modify the effect of PM2.5 on mortality.

Our quantitative analysis modeled primary PM2.5 emissions and did not include the influence of reducing NOx

emissions on secondary ammonium nitrate (particle) formation. This omission is likely to underestimate the

potential health benefits of road pricing.

It is important to note that bicycle commuters receive relatively high doses of air pollution compared with car

commuters, given increased time in traffic, proximity to sources, and elevated minutes of ventilation – though not

likely enough to counterbalance the physical activity benefits. 61 Increases in air pollution exposure associated

with active transportation were not considered in the analysis of air pollution effects.

Table 9. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects for Traffic‐related

PM2.5 and Premature Mortality

Factors Affecting Certainty Assessment Approach Summary

Confidence Level

Exposure Assessment Reliable small level population estimates. Method for parcel

level assignment not validated but defensible. Validated traffic

volume estimates. Assuming share of truck traffic remain

constant using EMFAC emissions models. Bus volumes not

included; would not anticipate significant changes in impacts

due to fleet conversions to alternative fuels. Emissions models

(EMFAC) valid in base and future years. High uncertainty with

precision of dispersion model in urban environment. Model

approach has been validated based on limited measured

values.

Moderate

Baseline Disease Prevalence Mortality Rate: County‐level data from vital statistics. High

Exposure‐Response Function

(ERF)

Modest variation in ERF in the most reliable studies. ERF not

specific to San Francisco but applicable in exposure range

according to CARB expert consensus. ERF does not take into

account differential vulnerability within the SF population.

Moderate





B. Road Traffic Noise

Health Effects of Road Traffic Noise

Roadway vehicles represent the primary source of urban environmental noise.62 Exposure to noise is a function of

the distance from roadways, vehicle volume, vehicle and engine type, operating speed, and roadway conditions.

Health effects of environmental noise, including roadway noise, depend on the intensity of noise, on the duration

of exposure, and the context of exposure. While individual sensitivity to noise may vary substantially, moderate

levels of noise can limit or interfere with the ability to conduct daily tasks and activity— to have an ordinary

conversation, enjoy a leisure activities, rest, sleep, concentrate or get tasks done. Noise annoyance is defined as “a

feeling of resentment, displeasure, discomfort, dissatisfaction, or offense when noise interferes with someone's

thoughts, feelings, or actual activities.”63

Sufficient scientific evidence documents that chronic exposure to moderate levels of noise below levels required

for mechanical damage to hearing can result in other health and physiological impacts including cognitive

impairment, decreased school performance, sleep disturbance, and hypertension and ischemic heart disease.64

Numerous well designed studies also show that children exposed to chronic transportation noise have deficits in

school performance and educational outcomes.65

As a physiological stressor that affects the autonomic nervous system and the endocrine system, noise may harm

the cardiovascular system. Substantial emerging evidence links moderate levels of traffic noise (>65 dB(A)) to

hypertension and ischemic heart disease.66 In a meta‐analysis of 43 studies of noise exposure and heart disease

published in 2002, road traffic noise was associated with higher risk for myocardial infarction and ischemic heart

disease than in the general population, and air traffic noise was associated with consultation with a doctor about

heart problems, use of cardiovascular medications, and angina pectoralis.67 In the same meta‐analysis, exposure

to air traffic noise was statistically associated with hypertension with an estimated relative risks per 5 dB(A) noise

increase of 1.26 (1.14‐1.39). In a 2009 meta‐analysis, Babisch synthesized a dose‐response function between

increasing noise levels above 60 dB(A) and cardiovascular risk.68 A large study published in 2009 found a notable

effect of noise on hypertension at > 64 dB(A) (OR 1.45, 95% CI 1.04 ‐ 2.02) with age acting as an effect modifier

(effects in middle aged 40‐59).69

Figure 12 illustrates the causal pathways through which road pricing policy may impact health via changes in

traffic‐related noise. Similar to the pathway for air pollutant impacts, road pricing increases the cost and therefore

reduces the demand for auto trips. Consequentially, pricing could decrease local traffic volumes, resulting in less

traffic‐related noise and diminishing its resulting health impacts. Changes in the composition of traffic based on

vehicle type (e.g., increases in buses) may counterbalance reductions in noise associated with reduced traffic

volumes. Increases in speed could also contribute to increased noise.



Figure 12. Pathway from Road Pricing Policy to Traffic‐Related Noise Health Impacts

Analysis of Traffic‐Attributable Noise Levels and Associated Health Impacts

As discussed above, traffic noise is causally associated with numerous effects on health, and changes in the

population exposure to traffic noise would be expected to result in corresponding changes to population health.

To evaluate the magnitude of the effect, we first modeled how traffic‐attributable noise levels in San Francisco

would change given the alternative scenarios. We then calculated the effect of the changes in noise levels on two

outcomes—community annoyance and heart attacks. Analytic methods are described in Appendix C.

The maps in Figure 13 depict traffic‐attributable noise levels in three scenarios: 2005, 2015 BAU, and 2015 RP at

the residential parcel level. The highest noise levels are concentrated in and on the boundaries of the congestion

pricing zone area – with noise levels elevated near freeways and highways and increasing from 2005 (upper left) to

2015 conditions both under BAU (upper right) and with road pricing (lower left). These differences in noise levels

are predicted by streets with higher traffic volumes, slower congested speeds, and higher proportions of traffic

comprised of trucks and/or buses proximate to residential uses.

Figure 13. Traffic attributable modeled noise levels in 2005, 2015 BAU, and 2015 RP: Residential

Parcels in San Francisco, California





With regards to health, noise levels shown on the maps can be interpreted using thresholds or values for noise

levels compatible with human activities and health. For example, human speech can be understood readily at

background noise levels of 45 dBA, but humans require much more vocal effort to be understood when the

background noise level is about 65 dBA. Table 10 provides guidance values for health‐protective levels of noise in

specific environments. The World Health Organization developed these guidelines based on a review of critical

noise‐related health effects.70 The values in the table reflect desirable and health protective levels of noise;

however, it is important to note that they have been developed without regard to context or feasibility and are not

regulatory standards.

Table 10. World Health Organization Guideline Values for Community Noise in Specific Environments

(1999)

Specific environment

Critical Health effect(s) LAeq [dB(A)]

Time base [hours]

LAmax fast [dB]

Outdoor living area Serious annoyance, daytime and eveningModerate annoyance, daytime and evening

55 50

16 16

‐ ‐

Dwelling, indoors Inside bedrooms

Speech intelligibility & moderate annoyance, daytime & evening Sleep disturbance, night‐time

35 30

16 8

‐ 45

Outside bedrooms Sleep disturbance, window open (outdoor values)

45 8 60

School class rooms & preschools, indoors

Speech intelligibility, disturbance of information extraction, Message communication

35 during class

‐

Pre‐school bedrooms, indoor

Sleep disturbance 30 sleeping time

45

School, playground outdoors

Annoyance (external source) 55 during play

‐

Hospital, ward rooms, indoors

Sleep disturbance, night‐time Sleep disturbance, daytime and evenings

30 30

8 16

40

Hospital treatment rooms, indoors

Interference with rest and recovery (#1)

Industrial commercial shopping and traffic areas, indoors and outdoors

Hearing impairment 70 24 110

Ceremonies, festivals and entertainment events

Hearing impairment (patrons:<5times/year)

100 4 110

Public addresses, indoors and outdoors

Hearing impairment 85 1 110

Music and other Sounds through headphones/earphones

Hearing impairment (free‐field value) 85 (#4) 1 110

Impulse sounds from toys, fireworks and firearms

Hearing impairment (adults) Hearing impairment (children)

‐ ‐ 140 (#2) 120 (#2)

Outdoors in parkland and conservations areas

Disruption of tranquility #3

#1: As low as possible

#2: Peak sound pressure (not LAF, max) measured 100 mm from the ear

#3: Existing quiet outdoor areas should be preserved and the ratio of intruding noise to natural background sound should be

kept low



#4: Under headphones, adapted to free‐field values

Tables 11 a and b summarize the citywide changes in noise levels depicted at the parcel level in the above maps.

Table 11a summarizes the changes from 2005 to 2015 BAU, 2% of the residential population living in areas with

noise levels below 55 dBA under 2005 transportation conditions are expected to experience an increase in traffic‐

related noise exceeding 55 dBA (based on the day‐night average sound level, Ldn) – the threshold for noise‐

related annoyance. An additional 5% of the population already living in areas exceeding 55 dBA in 2005 conditions

could experience a substantial increase in noise levels. No groups are expected to experience a decrease in noise

levels from 2005 – 2015 BAU conditions.

Table 11b summarizes the difference among noise levels between 2015 BAU to 2015 Road Pricing conditions.

Under 2015 RP, 1% of the residential population living in areas with 2015 BAU noise levels below 55 dBA are

expected to experience an increase in traffic‐related noise to exceed 55 dBA. In addition, 6% of the population

already living in areas exceeding 55 dBA in 2015 conditions experience significant increases in noise levels that

shift them to a higher noise category. However, the 2015 RP scenario also appears to have a moderating effect on

noise levels for some residents relative to 2015 BAU conditions – with 5% of residents experiencing a decrease of

community noise levels with road pricing.

Table 11a. Estimated Traffic Attributable Noise (Ldn) Levels and 2015 Residential Population

Exposures in 2005‐2015 BAU: San Francisco, California (n=824,000)

Noise Change Compared to 2005 0‐50 51‐55 56‐60 61‐65 66‐70 71‐87 Total

Stayed the same 17% 8% 14% 17% 17% 19% 91%

Changed but stayed quiet (<55 ldn) in 2015 BAU 0% 1% 0% 0% 0% 0% 1%

Quiet (<55 ldn) and got louder (>55 ldn) in 2015 BAU 0% 0% 2% 0% 0% 0% 2%

Loud (>55 ldn) and got quieter with BAU 0% 0% 0% 0% 0% 0% 0%Loud (>55 ldn) and got louder in 2015 BAU 0% 0% 0% 2% 2% 1% 5%

2015 BAU Noise Level in Decibels (Ldn)

Table 11b. Estimated Traffic Attributable Noise (Ldn) Levels and 2015 Residential Population

Exposures in 2015 BAU‐2015 RP: San Francisco, California (n=824,000)

Noise Change Compared to 2015 BAU 0‐50 51‐55 56‐60 61‐65 66‐70 71‐87 Total

Stayed the same 13% 8% 12% 16% 17% 17% 83%Changed but stayed quiet (<55 ldn) in 2015 RP 4% 2% 0% 0% 0% 0% 6%Quiet (<55 ldn) and got louder (>55 ldn) in 2015 RP 0% 0% 1% 0% 0% 0% 1%Loud (>55 ldn) and got quieter with 2015 RP 0% 0% 1% 1% 1% 2% 5%Loud (>55 ldn) and got louder in 2015 RP 0% 0% 2% 2% 1% 1% 6%

2015 RP Noise Level in Decibels (Ldn)

We used these estimated changes in population exposure to traffic‐related spatial and temporal noise levels to

estimate changes in expected perceptions of annoyance and ischemic heart disease events using concentration‐

response functions from empirical research (detailed methods in Appendix C).71 The key findings are provided in

Table 12.





Table 12. Noise‐related Health Effects in 2005, 2015 Business As Usual, and 2015 with Road Pricing


Health Impacts

(Annual Estimates)

2005 2015 BAU 2015 RP Change:

2005 →

2015 BAU

Change:

2005 →

2015 RP

Change: 2015

BAU → 2015

RPNoise, High Annoyance (N, Ages 25 and up)

Citywide 92,500 100,400 100,300 7,900 7,800 ‐100Congestion Pricing Zone 36,800 40,600 40,500 3,800 3,700 ‐100

Noise, Myocardial Infarction Associated with Traffic Noise (N, Ages 30 and up)Citywide 31 34 34 3 3 0


Citywide, for 2015 BAU relative to 2005, we estimate a 9% increase in the number of people age 25 or older who

would experience perceived high annoyance (an additional 7,900 people highly annoyed); an additional three

heart attacks annually attributable to traffic noise are estimated. There was no substantial difference in these

effects when we looked only within the cordon zone: there would be an estimated 10% increase in number of

people affected by annoyance, and two of the three heart attacks would occur in the cordon area.

Citywide, for 2015 RP relative to 2015 BAU, the analysis estimated a slight decrease in perceived high annoyance

(<1%, 100 fewer people highly annoyed) and no change in myocardial infarction attributable to traffic noise. Based

on the cordon zone analyses, we see the very slight decrease in perceived high annoyance was estimated in the

cordon area.

Health Effects Characterization: Noise‐related health effects


annoyance and myocardial infarction predicted by traffic‐related noise. Based on the available evidence: it is very

likely that the proposed policy will impact traffic‐related noise effects on annoyance, and possible/likely that the

policy will affect noise‐related myocardial infarction; severity of the effect is low for annoyance and high for

myocardial infarction ; the magnitude of the predicted change is limited for both outcomes relative to 2015 BAU;

and the health effects particularly in the cordon zone contribute to a restorative equity effect.



Table 13. Road Pricing Policy Health Effects Characterization: Traffic‐related Noise Health Effects

Health Effects:

Characteristics






High for Annoyance; Moderate for

Myocardial Infarction


affect traffic‐noise related annoyance and myocardial

infarction, irrespective of the frequency, severity, or magnitude

of the effect?

Very Likely/Certain for Annoyance:

Consistent evidence for causality from

epidemiologic studies with diverse

populations, noise levels, and exposure

ranges. Possible/Likely for Myocardial

Infarction (MI): Plausible evidence for

causal relationship between noise and MI

prevalence with studies supporting

physiological effects but limited

epidemiology.

Severity How important is the effect on traffic‐noise related annoyance

and myocardial infarction with regards to human function, well‐

being or longevity, considering the affected community's

current ability to manage the health effects?

Medium: Acute, chronic, or permanent

effects that substantially affect function,

well‐being, or livelihood but are largely

manageable within the capacity of the

community health system.

Magnitude How much will traffic‐noise related annoyance and myocardial

infarction change as a result of the road pricing policy under

study?

Limited to Moderate (High Certainty for

Annoyance; Moderate Certainty for MI):a


very modestly moderates increases in

annoyance in the congestion pricing area;

there is not evidence of an impact on

myocardial infarction.




Restorative Equity Effects: Populations

in the congestion pricing area have been

historically, disproportionately burdened

with traffic‐related noise in their

communities. The road pricing policy

under study has the potential to reduce

some of this inequity and related adverse

health effects.

a See Table 14 for a summary of factors impacting on the certainty of the magnitude of the health effects

characterization.

Certainty in Health Effects Characterization

As detailed in the introduction to this section, there exists a high level of causal certainty for the effects of noise on

annoyance. Studies of the effect of noise on community annoyance are consistent across international contexts. A

causal relationship between noise and myocardial infarction (heart attack) is less well established though research

studies support the biological plausibility of this effect. Epidemiology on the subject is consistent but is limited and

therefore reduced our degree of confidence in the likelihood of this effect.

We have high‐moderate confidence in our exposure analysis; noise levels estimated in the baseline case conform

closely to actual field measures. We conservatively estimated that truck volumes would not change in 2015 BAU





vs. 2015 RP conditions. The exposure can be assigned to residential parcels and the population enumeration is

reliable.

Our model used a standard bus noise level estimate and thus did not account for differences in noise levels based

on bus vehicle type (e.g., electric trolley vs. diesel vs. hybrid buses). The San Francisco Municipal Transportation

Agency (SFMTA) intends to replace diesel buses with quieter hybrid buses in its bus fleet. Compared to diesel

buses used for San Francisco's public transit system, noise levels produced by new hybrid buses are up to 10

decibels less than current vehicles based on data provided by the SFMTA. These decreases could contribute to

notable reductions in population noise exposure estimates and perceived noise for residents living within 25

meters of bus routes in both 2015 BAU and 2015 RP future scenarios. We had high confidence in the exposure‐

response functions for perceived noise annoyance. As detailed in Appendix C, a meta‐analysis conducted for this

outcome combined data from multiple international studies producing a robust exposure‐response function that

was applied in the exposure range of the levels reported in this HIA. The precision of the exposure response

function for noise and myocardial infarction has some uncertainty because the meta‐analysis on which it was

based included a small number of studies.

Table 14. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects of Traffic‐related

Noise


Confidence Level

Exposure Assessment Reliable small level population estimates. Method for parcel

level assignment not validated but defensible. Model limited to

traffic‐related noise sources yet validated (correlated) against

field noise measures. Potentially more uncertainty in future

truck and bus traffic noise estimates. Uncertainty from noise

emissions control technologies, including quieter hybrid bus

engines.

High‐Moderate

Baseline Disease Prevalence Baseline prevalence of noise annoyance unavailable and thus

estimated. Local disease prevalence data on myocardial

infarction (MI) available though hospital discharge data may

underestimate MI prevalence.

Moderate


(ERF)

Community annoyance ERFs are specific to road noise sources

and represents pooled analysis, multiple studies, diverse

contexts, though not calibrated for San Francisco populations.

MI ERF based on limited studies.

High (Community

annoyance);

Moderate

(Myocardial

infarction)



C. Active Transportation

Health Effects of Physical Activity

Evidence that physical activity has multiple health benefits is unequivocal. A 2008 comprehensive review

documents the particularly strong evidence for a causal relationship between activity level and enhanced

cardiorespiratory and muscular fitness, cardiovascular and metabolic health biomarkers, bone health, body mass

and composition in children and youth. In adults and older adults, strong evidence demonstrates that, compared

to less active counterparts, more active men and women have lower rates of all‐cause mortality, coronary heart

disease, high blood pressure, stroke, type 2 diabetes, metabolic syndrome, colon cancer, breast cancer, and

depression. For older adults, strong evidence indicates that being physically active is associated with higher levels

of functional health, a lower risk of falling, and better cognitive function.” This research reported reasonably

consistent findings specifically for the health benefits of walking – showing a consistently lower risk of all‐cause

mortality for those who walk 2 or more hours per week.72 A 2011 report issued by an international group of

experts of data from Copenhagen documents similar all‐cause mortality benefits from regular cycling for

commuting controlling for socio‐demographic and leisure time physical activity.73 While regular physical activity

can help people lead longer, healthier lives, a 2009 summary by the Robert Wood Johnson Active Living Research

program revealed that fewer than 50% of children and adolescents and fewer than 10% of adults in the U.S.

achieve public health recommendations of 30 to 60 minutes per day of moderate‐ to vigorous‐intensity physical

activity on 5 or more days of the week recommendations.74

There are multiple environmental barriers that both children and adults face to achieving recommended levels of

physical activity, including: limited discretionary time; barriers to accessing parks and recreational areas;

reductions in school physical education programs; and sidewalks, streets, or outdoor spaces that are not or are

not perceived as safe to use. Encouraging and facilitating active transportation – walking or cycling as a form of

travel for utilitarian trips – is a key strategy for increasing daily physical activity. 75 Built environmental factors that

are associated with active transportation via walking and cycling include increased resident and employment

density, greater diversity of land use mix (e.g., residential land use near retail land uses), shorter distances

destinations, and street design factors (e.g., grid street networks, the presence of sidewalks).76

Based on the evidence above, Figure 14 details a pathway through which road pricing policy may affect health via

active transportation. Simply stated, as the costs of auto trips increase active transportation is incentivized as a

free alternative resulting in a greater number of trips made by walking or biking and therefore increasing physical

activity and its associated health benefits. Additionally, road pricing generates revenue that may be invested in the

transportation system to improve environments for walking and cycling and further increase trip‐making.

Figure 14. Pathway from Road Pricing Policy to Physical Activity via Active Transportation





Analysis of Active Transportation and Lives Saved from Walking and Cycling

We used existing and future conditions data from the SFCTA to assess walk and bike trips and trip times for trips

made by residents living in 12 geographic districts in San Francisco (the smallest reliable level of aggregation used

to estimate walk and bike trips in the SFCTA’s transportation analyses). We were specifically interested in

estimating how changes in active transportation varied by age. We had data for existing and future conditions for

area‐level resident population by age, but because data was not available on trip making and travel mode by age in

San Francisco, we estimated the age distribution of trips by transportation mode based on the Bay Area Travel

Survey data, assuming the proportion making weekday trips and transportation mode by age were similar across

the Bay Area region.

As depicted in the map in Figure 15, we collapsed the 12 districts into three districts for this analysis, defined with

respect to their geographic location relative to the Northeast Cordon road pricing zone (depicted with a blue

boundary on the map). The three zones are “In the Zone,” turquoise, comprised of the districts mostly in the

cordon zone; “On the Fringe,” yellow/mustard, comprised of the districts just outside of the cordon zone; and the

“Outer Districts,” pink, representing those farther away from the cordon zone.

Figure 15. Activity Zones used for Active Transportation Analyses and Proportion of Residents in Each

Zone by Age: San Francisco, California

Age

In the Cordon

On the Fringe

Outer Districts

In the Cordon

On the Fringe

Outer Districts

Under 5 22% 22% 57% 23% 21% 56%5-19 19% 19% 62% 19% 18% 63%

20-44 36% 23% 41% 37% 22% 41%45-64 31% 19% 50% 31% 19% 50%

65 and Over 35% 14% 52% 35% 14% 51%Total 32% 20% 48% 32% 20% 48%

2005 2015

Age

In the Cordon

On the Fringe

Outer Districts

In the Cordon

On the Fringe

Outer Districts

Under 5 22% 22% 57% 23% 21% 56%5-19 19% 19% 62% 19% 18% 63%

20-44 36% 23% 41% 37% 22% 41%45-64 31% 19% 50% 31% 19% 50%

65 and Over 35% 14% 52% 35% 14% 51%Total 32% 20% 48% 32% 20% 48%

2005 2015

Table 15 summarizes the predicted changes in walking under the different study scenarios.



Table 15. Walk Trips and Walk Minutes Per Capita in 2005, Percent Change from 2005 to 2015

“Business As Usual” (BAU), and Percent Change from 2015 BAU to 2015 with Road Pricing (RP): San

Francisco, California

Walk Trips Per

Capita

Walk Minutes

Per Capita

% Change in

Walk Trips Per

Capita

% Change in

Walk Minutes

Per Capita

% Change in

Walk Trips Per

Capita

% Change in

Walk Minutes

Per Capita

In The Zone

Age Under 5 1.6 30 6% 7%

Age 5‐19 2.5 46 6% 7%

Age 20‐44 1.8 32 5% 7%

Age 45‐64 1.1 20 5% 6%

Age 65 and Over 2.0 36 5% 6%

TOTAL 1.7 31 5% 7% 2% 2%

On the Fringe

Age Under 5 0.9 18 0% 1%

Age 5‐19 1.3 27 0% 1%

Age 20‐44 1.0 21 0% 1%Age 45‐64 0.6 13 0% 1%

Age 65 and Over 1.0 22 0% 0%

TOTAL 0.9 19 0% 1% 3% 3

Outer Districts

Age Under 5 0.6 11 ‐1% 0%

Age 5‐19 0.9 17 ‐1% 0%

Age 20‐44 0.6 13 ‐1% 0%

Age 45‐64 0.4 8 ‐1% 0%

Age 65 and Over 0.7 14 ‐1% ‐1%

TOTAL 0.6 12 ‐1% 0% 2% 1%

2005 2005 ‐ 2015 BAU 2015 BAU ‐ 2015 RP

%

Note: percentages are listed only for the TOTAL in the 2015 BAU – 2015 RP scenario because the percentage are

constant across age groups given the assumptions used to estimate the changes.

2005: In 2005 conditions, across all age groups, residents are walking more “In the Zone” – almost double that of

residents “On the Fringe” and approximately three times that of residents of the Outer Districts. This is consistent

with increased residential density, decreased parking availability and lower household car ownership, greater

transit access and greater household proximity to goods and services “In the Zone” relative to the other districts –

all factors supportive of increased walking.

2005 → 2015 BAU: Under 2015 BAU compared to 2005, there are notable increases in walk trips and time per

capita only “In the Zone.” This can be explained by projected increases in number of residents and existing land

use and transportation characteristics supportive of walking. Walk trips may also increase due to increases in

vehicle congestion and expected increases in public transit costs.

2015 BAU →2015 RP: Under 2015 with RP compared to 2015 BAU, there are modest increases in walk trips across

all districts – including a 3% increase “On the Fringe.” Residents with short distances trips through the road pricing

boundary may switch modes from vehicles to walking when the cost of driving increases.





Table 16 summarizes the predicted changes in biking under the different study scenarios.

Table 16. Bike Trips and Bike Minutes Per Capita in 2005, Percent Change from 2005 to 2015 “Business


California

Bike Trips Per

Capita

Bike Minutes

Per Capita

% Change in

Bike Trips Per

Capita

% Change in

Bike Minutes

Per Capita

% Change in

Bike Trips Per

Capita

% Change in

Bike Minutes

Per Capita

In The Zone

Age Under 5 0.0 0.4 9% 11%

Age 5‐19 0.3 2.7 10% 12%

Age 20‐44 0.2 2.6 9% 10%

Age 45‐64 0.1 0.8 8% 10%

Age 65 and Over 0.1 0.6 7% 9%

TOTAL 0.2 1.8 7% 9% ‐2% ‐1%

On the Fringe

Age Under 5 0.0 0.5 7% 6%

Age 5‐19 0.2 3.2 6% 6%

Age 20‐44 0.2 3.1 7% 7%

Age 45‐64 0.1 1.0 7% 7%

Age 65 and Over 0.0 0.7 6% 6%

TOTAL 0.2 2.2 5% 5% 5% 4%

Outer Districts

Age Under 5 0.0 0.5 4% 6%

Age 5‐19 0.2 2.9 4% 6%

Age 20‐44 0.1 2.7 3% 6%

Age 45‐64 0.0 0.9 3% 5%

Age 65 and Over 0.0 0.6 3% 5%

TOTAL 0.1 1.8 2% 4% 3% 3%

2005 2005 ‐ 2015 BAU 2015 BAU ‐ 2015 RP

Note: percentages are listed only for the TOTAL in the 2015 BAU – 2015 RP scenario because the percentages are

constant across age groups given the assumptions used to estimate the changes.

2005: In 2005 we see relatively low bike trips and time travelling per capita across the zones compared to walking.

2005 → 2015 BAU: Under 2015 BAU compared to 2005 we see notable predicted increases in bike trips per capita

across the zones. This is consistent with recent observed increases in biking and related investments in bicycling

infrastructure. Relatively higher increases in the zone and the fringe may be explained by environments which are

supportive of bike trips including bike lanes and relatively flatter terrain (for San Francisco).

2015 BAU →2015 RP: Under 2015 RP compared 2015 BAU we estimate additional increases in bike trips “On the

Fringe” and in the “Outer Districts” As the cost of driving increases some residents are expected to switch modes

from car to bike for trips across the cordon. Per capita bike trips slightly decrease “In the Zone” under 2015 with

road pricing. This may be due, in part, to increases in transit trips replacing bike trips as road pricing revenue

investments improve transit service in that area Despite this slight predicted decrease in biking “In the Zone,”

aggregate active transportation via biking and walking increases in all areas under 2015 RP compared to 2015 BAU.



Health Effects of Increases in Active Transportation

Physical activity has numerous health benefits, as summarized in the beginning of this section. Change in life

expectancy is one summative outcome that captures multiple effects of physical activity on human disease and

well‐being.

To quantify the health benefits from increases in walking and cycling, we used the HEAT (Health Economic

Assessment Tool) for Walking and Cycling developed by the World Health Organization’s Regional Office for

Europe (see Appendix D for further description of this tool and our approach). The HEAT tool for walking is based

on a meta‐analysis of international studies that found a walking exposure of 29 minutes seven days a week results

in a 22% reduction in the risk of pre‐mature mortality (Relative Risk = 0.78, 0.64‐0.98 95% confidence interval). The

HEAT relative risk estimate for all‐cause mortality for cycling was estimated to be 0.72 based on a meta‐analysis of

three cohort studies from Copenhagen. These studies compared commuter cyclists aged 20‐60 years to the

general population, controlling for socioeconomic variables (age, sex, smoking etc.) as well as for leisure time

physical activity.77 As described earlier in this section, the SFCTA travel model provided data on walking and

cycling trip and duration by residents at the 12‐district level under all scenarios, and we further estimated the age

distribution of trips by transportation mode based on the Bay Area Travel Survey data. The effect on mortality

incidence was applied to locally specific mortality rates to estimate lives saved from walking and cycling.

Walking: As detailed in Table 17, we found that the modest predicted increases in walking could save an

additional eight lives each year under 2015 “business as usual” compared to existing conditions, and three

additional lives each year beyond BAU with 2015 RP.

Cycling: As summarized in Table 17, we found that the modest predicted increases in cycling could save an

additional two lives each year under 2015 “business as usual” compared to existing conditions, and one additional

life each year beyond BAU with 2015 RP.

Table 17. Lives Saved from Active Transportation in 2005, 2015 Business As Usual, and 2015 with


Health Impacts

(Annual Estimates)

2005 2015 BAU 2015 RP Change:

2005 →

2015 BAU

Change:

2005 →

2015 RP

Change: 2015

BAU → 2015

RPCycling for Active Transportation, Lives Saved (N, Ages 25‐64)

Citywide 23 25 26 2 3 1Congestion Pricing Zone 8 9 9 1 1 0

Walking for Active Transportation, Lives Saved (N, Ages 25‐64)Citywide 130 138 141 8 11 3


Health Effects Characterization: Lives Saved from Active Transportation Via Walking and Cycling

Table 18 summarizes our overall characterization of the health effects of the road pricing policy related to active

transportation and the associated lives saved. Based on the available evidence: it is very likely that the proposed

policy will increase active transportation and associated lives saved; severity of the effect— including acute and

chronic respiratory diseases and mortality are high; the magnitude of the predicted decrease in premature deaths

is limited relative to 2015 BAU; and the reduction of deaths particularly in the cordon zone contributes to a

restorative equity effect.





Table 18. Road Pricing Policy Health Effects Characterization: Lives Saved from Active Transportation

via Walking and Cycling

Health Effects:

Characteristics






Moderate


affect active transportation and associated lives saved via

increases in walking and biking, irrespective of the frequency,

severity, or magnitude of the effect?

Very Likely/Certain: Adequate evidence

exists for a causal and generalizable

effect based on the empirical literature,

including meta‐analyses and reviews

conducted by a team of international

experts including the World Health

Organization. Evaluation studies of

congestion pricing in other countries

support increases in active

transportation.

Severity How important is the effect with regards to human function,

well‐being or longevity, considering the affected community's

current ability to manage the health effects?

High: The impacts on active

transportation are associated with

reductions in mortality.

Magnitude How much will lives saved due to increases in walking and

biking change as a result of the road pricing policy under study?Limited (Moderate Certainty):a


contributes to an additional 4 lives saved

attributable to walking and cycling

annually in San Francisco.




Insufficient Evidence to Evaluate: Data

not available at the refined level, in terms

of area or individual characteristics,

needed to evaluate the distibution of

effects.

a See Table 19 for a summary of factors impacting on the certainty of the magnitude of the health effects

characterization.

Certainty in Health Effects Characterization

The SFCTA’s model predicted modest increases in active transportation citywide in both future scenarios but

greater increases with road pricing. Congestion pricing evaluations in other countries have found similar increases

in active transportation.78

Because trip making by age group is not estimated by the SFCTA model, we assumed that trip making and

transportation mode by age in San Francisco were similar to that of the Bay Area region in order to use Bay Area

Travel Survey data. The model currently does not include leisure trips, and is likely an underestimate of walking

and cycling, overall. Our estimate also does not include walking trips to transit – and thus likely underestimates

the overall health benefits of walking for transportation in all scenarios.

As discussed above, we estimated lives saved from changes in walking and cycling using the World Health

Organization’s health economic assessment tool (HEAT). HEAT provides an effect measure (exposure‐response

relationship) for walking and cycling specifically (versus physically active time more generally). The exposure‐

response relationship for walking is based on meta‐analysis of international studies controlled for numerous



confounders. The relative risk estimate for cycling is based on three studies in a European population, and thus

there is some uncertainty about the generalizability of the estimate to a U.S. population. A meta‐analysis

published in 2011 of physical activity and mortality supports that the size of the protective effect for daily walking

or bicycling on mortality in the HEAT tool is similar to the benefit of meeting the U.S. Surgeon Generals

recommendations for adult physical activity.79 The developers of the HEAT approach acknowledge that the focus

of the HEAT tool on mortality only (and not morbidity) and adult populations ‐ where there is currently the

strongest body of evidence regarding health benefits ‐ likely produces very conservative estimates of health

benefits of active transportation in the population.80

The HEAT estimation approach treats exposure to walking as a dichotomous variable and predicts a health benefit

(reduction in mortality) for specific population of individuals meeting a specific threshold of walking or cycling

activity during a specified time period. There is good evidence for an inverse linear relationship between the

amount of physical activity and mortality. In other words, every additional amount of physical activity has a

proportional benefit to health. However, others have found that the benefit of increasing physical activity varies at

different levels of physical activity.81 Assuming that the inverse‐linear relationship is the more accurate

representation of the actual biologic processes, estimating benefits using a threshold approach could

underestimate the benefits of changes in active transportation.

Table 19. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects for Lives Saved

from Active Transportation via Walking and Cycling


Confidence Level

Exposure Assessment Changes in walking and cycling trips and time for

transportation based on model outputs from the

SFCTA model at a district level, the smallest area

level with reliable estimates. Used BATS (2000) data

to estimate trips by age, assuming the proportion of

the population travelling during the average

weekday and average number of daily trips are the

same within San Francisco as the rest of the Bay

Area. Does not include leisure walking and cycling

trips, or walking trips to transit. Likely an

underestimate of walking and cycling, overall.

Moderate

Baseline Disease Prevalence Mortality Rate: County‐level data from vital statistics

for people aged 25‐64.

High


(ERF)

Health Economic Assessment Tool (HEAT) for

walking and cycling approach. Estimate for walking

based on meta‐analysis of nine studies. Estimate for

cycling based on three studies and expert consensus.

A threshold approach, though evidence for an

inverse‐linear relationship, which could result in an

underestimate of overall benefits. Overall health

benefits underestimated given tool focus on adults,

mortality only.

Moderate





D. Vehicle-Pedestrian Injury Collisions

Road Pricing Effects on Pedestrian Safety

Traffic injuries and fatalities represent a significant and avoidable health burden with 33,808 deaths and 2,217,000

injuries in the United States alone in 2009. Among transportation system users pedestrians and pedal‐cyclists are

arguably the most vulnerable users. In the U.S., pedestrians, who are more likely to be lower‐income and not own

automobiles, experience a disproportionate share of traffic injuries and fatalities risks than vehicle occupants on

both a per‐trip and per‐mile basis with 4,092 pedestrian deaths in 2009.82

Pedestrian and traffic volumes, vehicle speed, roadway width, intersection design and geometry, land use,

pedestrian and driver behavior, and vehicle design are all factors that determine the frequency of pedestrian‐

vehicle collisions.83 Vehicle and pedestrian volumes are the most important single determinants of pedestrian

injury collision frequency, with non‐linear but positive associations between pedestrian flows and traffic volumes

with pedestrian injury collisions consistently found at the small‐area level including intersections.84 Studies also

show heterogeneity in these associations, potentially due to differences in other environmental conditions

including land use, transportation system, and socio‐demographic factors. Roadway design and higher operating

speeds are potent determinants of injury frequency as well as injury severity. Quality of pedestrian facilities are an

important predictor of pedestrian injury in the research literature, based on analyses at the intersection, street

segment, and small‐areas levels.85 Low‐income neighborhoods are more heavily burdened with pedestrian injuries

and fatalities due factors potentially including higher residential area traffic densities, greater use of active

transport and public transit, and relatively poorer quality roadway facilities.86

The objective of road pricing policy is to reduce traffic congestion on local roadways during peak use periods,

which also correspond to times of high pedestrian volumes. Reductions in traffic volume could reduce pedestrian

injury. However, shifts to active transportation and public transportation (which involves active transportation)

also simultaneously increase the number of pedestrians exposed to traffic hazards. Where free flow conditions do

not currently exist, reductions in traffic volume may also increase traffic speed which is a risk factor for collision

frequency and severity. Investments in road infrastructure funded by road pricing could address gaps in

pedestrian safety engineering countermeasures and mitigate potential adverse impacts. Figure 16 details a

pathway through which road pricing policy impacts changes in vehicle trips and therefore local traffic volumes,

traffic hazards and ultimately vehicle‐pedestrian injury collisions.

Figure 16. Pathway from Road Pricing Policy to Vehicle‐Pedestrian Injury



Analysis of Vehicle‐Pedestrian Injury Collisions

For this analysis, we estimated changes in pedestrian injury collision frequency (absolute) numbers in San

Francisco census tracts using an area‐level vehicle‐pedestrian injury forecasting model developed by SFDPH.87 The

model was developed to predict changes in injury collisions attributable to changes in traffic volumes and

population generally assuming no changes in roadway infrastructure (further detailed in Appendix E). Traffic

volume was the single strongest predictor of pedestrian injury collisions in the model.

We applied traffic and population estimates to the model to estimate pedestrian‐injury collisions under each

scenario. Census tract‐level traffic volume varies in all three scenarios. Model parameters that vary only between

existing and future conditions are the number of residents and employees, the proportion of the population aged

65 and over, and proportion of the population living in poverty. Resident and employee populations serve as a

rough proxy for pedestrian volumes. While the model considers street type, the model does not account for

changes in traffic speeds or intersection level counter‐measures.

Table 20 summarizes the predicted changes in vehicle‐pedestrian injury collisions citywide as well as in the cordon

zone.

Table 20. Annual Estimated Pedestrian Injury Collisions in 2005, 2015 Business As Usual, and 2015


Health Impacts

(Annual Estimates)

2005 2015 BAU 2015 RP Change:

2005 →

2015 BAU

Change:

2005 →

2015 RP

Change: 2015

BAU → 2015

RPVehicle‐Pedestrian Injury Collisions (N, Total)

Citywide 810 860 815 50 5 ‐45Congestion Pricing Zone 360 395 360 35 0 ‐35

2005: Under existing (2005) conditions, we estimated 810 vehicle‐pedestrian injury collisions citywide. (Note: As

detailed in Appendix E, we estimated census‐tract level collisions for 2005 baseline conditions based on the

average of actual reported collisions for a five‐year period in an effort to reduce statistical anomalies. Police‐

reported pedestrian‐vehicle injuries and deaths in San Francisco have varied between approximately 710‐910

injuries and deaths annually since 2001.) Using the population data reported in Table 3, we calculated a

population rate of approximately 102/100,000 resident population – over five times the Healthy People 2010

target rate of <20/100,000 population.88 Approximately 44% of San Francisco’s vehicle‐pedestrian injury collisions

occur in the cordon zone that is under study, with a population rate of approximately 210/100,000 population –

over two times that of the citywide population rate. The potential congestion pricing zone has the highest density

of vehicle‐pedestrian injuries in the city – largely explained by relatively higher concentrations of traffic, resident

and employee populations.89 Figure 17a illustrates how the pattern of injury varies widely across San Francisco

census tracts.

2005 → 2015 BAU: In 2015 BAU, the model estimates that pedestrian injury collisions increase 6% citywide to

approximately 860 annually (104/100,000 population) from 810, and increase in the cordon zone by 10% to

approximately 395 annually (216/100,000 population) up from 360 in 2005 conditions. In this scenario, 46% of

vehicle‐pedestrian injury collisions occur in the cordon zone –increasing the citywide geographic disparities. These

increases are largely predicted by increases in traffic, as well as by the planned population and employment

growth (see Figures 7 and 8, Table 3).

2015 BAU → 2015 RP: The model estimates that pedestrian injury collisions decrease in the 2015 scenario with

Road Pricing compared to 2015 BAU. This is explained by decreases in aggregate traffic volume citywide and in the





cordon zone (Table 4, ‐4% and ‐11%, respectively). These decreases could result in pedestrian injury collision totals

approximately at 2005 counts, with estimated collision rates below 2005 levels (citywide rate: 99/100,000

population; cordon zone rate: 197/100,000).

The following maps (Figures 17 and 18) depict the estimated census tract‐level changes in the annual average

absolute number of vehicle‐pedestrian injury collisions in 2005, 2015 BAU and 2015 RP conditions. The maps show

increases in collisions under BAU – with increases most notable in areas in the road pricing zone as well as some

tracts south of the zone near the freeway. In 2015 RP, reductions in the road pricing zone relative to 2015 BAU

conditions are estimated– again, explained by reductions in traffic volumes in those areas. There are a few census

tracts, indicated in a salmon color, where there are modest estimated increases in vehicle‐pedestrian injury

collisions under 2015 RP predicted by increases in traffic volume in those tracts (as depicted in Figure 9, Map D).

Note that the highest category reported in the maps for the annual average number of collisions changes under

2015 BAU conditions due to increases in total census tract collision numbers. As noted earlier, these change

estimates do not account for estimated census tract changes in pedestrian volumes; implications for the results

are discussed later in the section in the discussion regarding uncertainty.

Figure 17a. Vehicle‐Pedestrian Injury Collisions in 2005 and 2015 “Business As Usual” (BAU)


Figure 17b. Estimated Change in Vehicle‐Pedestrian Injury Collisions from 2005 → 2015 “Business As

Usual” (BAU) Conditions: San Francisco, California



Figure 18a. Vehicle‐Pedestrian Injury Collisions in 2015 “Business As Usual” (BAU) and 2015 with Road

Pricing (RP) Conditions: San Francisco, California

Figure 18b. Estimated Changes in Vehicle‐Pedestrian Injury Collisions in 2015 “Business As Usual”

(BAU) → 2015 with Road Pricing (RP) Conditions: San Francisco, California

Because the congestion pricing scenario largely reduces traffic volumes during peak commute periods, we

considered the time of day of pedestrian injury collisions in San Francisco for the five‐year period 2004‐ 2008. The

blue line in Figure 19 indicates the annual average number of pedestrian injury collisions within the cordon zone

under study, the pink line the area outside the cordon zone, and the green line the overall citywide pattern. There

are evident peaks during the AM and PM peak periods, providing additional support regarding the plausibility of

traffic reductions during the peak commute times as contributing to reductions in pedestrian injury collisions.

Specifically, over half of pedestrian injury collisions during 2004‐2008 occurred during the AM and PM commute

periods that are the focus of the congestion pricing charge (53% of the total in the cordon zone, 61% of the total

outside of the cordon zone). The temporal pattern of collisions is not accounted for in our model, and our

estimates could therefore underestimate the potential impact of congestion pricing policy on pedestrian injury

given the policy reduces traffic volume during peak injury periods; conversely, there could also be changes in the

temporal distribution of pedestrian injury collisions as traffic shifts to non‐peak times.





Figure 19. Annual Average Pedestrian Injury Collisions by Time of Day (2004‐2008): San Francisco,

California

0

10

20

30

40

50

60

70

Early AM(3:01-

6:00am)

AM (6:01-9:00am)

Mid-Day(9:01am-3:30pm)

PM (3:31-6:30pm)

Late PM(6:31pm-3:00am)

Time Periods

In Cordon Zone

Outside Cordon Zone

Citywide

Additional Targeted Pedestrian Safety Analyses: High‐Injury Corridors and Pedestrian Environmental Quality

To provide additional data on existing conditions and risk factors related to pedestrian injuries and to inform our

HIA recommendations regarding pedestrian safety, we further analyzed high‐injury pedestrian corridors in San

Francisco, as well as the pedestrian environmental quality in communities with existing high numbers or predicted

increases in vehicle‐pedestrian injury collisions.

High‐Injury Corridors

Vehicle‐pedestrian injury collisions are often concentrated along corridors and in areas with an aggregation of

multiple environmental‐level risk factors including higher traffic volumes and greater pedestrian activity. Analysis

of police‐reported vehicle‐pedestrian injury collision data for 2005‐2009 underscores the concentration of injuries

in San Francisco.

The map in Figure 20 identifies high‐injury areas and corridors defined as closely‐spaced, high injury street

segments, each high‐injury street segment with an injury count >9. To give weight to injury severity, each severe

or fatal injury was counted as three non‐severe injuries. The 6.7% of city street miles identified as high‐injury

corridors (in blue) account for 55% of severe and fatal pedestrian injuries and 51% of total pedestrian injuries

during the analysis period. The patterns echo differences in census tract level injury frequency illustrated in

Figures 17 and 18.



Figure 20. High‐Injury Density Corridors for Vehicle‐Pedestrian Injuries (2005‐2009): San Francisco,

California

Pedestrian Environmental Quality in Areas with High Existing and Future Increases in Pedestrian Injury

We also collected baseline street and intersection conditions data for this HIA with the Pedestrian Environmental

Quality Index (PEQI). The PEQI is a quantitative observational instrument to describe and summarize street and

intersection environmental factors known to affect people's travel behaviors and pedestrian safety. The index

provides a total score for each street segment and intersection assessed, which is computed as the sum of 31

weighted intersection and street indictors. The scores are categorically ranked into five classifications ranging from

“ideal quality” for pedestrians to “poor quality” for pedestrians. The SFDPH developed the PEQI for a number of

purposes, including assessing existing physical conditions and demonstrating the need for improvements in the

course of land use and transportation planning, community education regarding built environmental factors that

either promote or discourage walking in local neighborhoods, and to fill a gap in data regarding the physical

environment on our city streets in order to assess how it predicts key health determinants such as physical activity.

The PEQI methodology is described in further detail in Appendix F.

We were particularly interested in assessing intersection and street level factors in the pedestrian environment

that support safe walking to inform our recommendations to improve pedestrian safety in and near the cordon

zone. As noted in Table 22, the SFDPH Vehicle‐Pedestrian Injury Collision Model does not account for these

intersection and street specific factors. For PEQI data collection, we focused on areas with existing high number of





injuries as well as areas with projected increases in injuries based on our vehicle‐pedestrian injury forecasting

model findings. PEQI scores reflect the degree to which environmental factors supportive of walking and

pedestrian safety has been incorporated into street segment and intersection design. The PEQI scores street

segments and intersections separately, on a scale from 0 ‐100. We are currently using the following categories for

scoring, a priori, with equal intervals of 20 points for all categories:

100‐81 = highest quality, many important pedestrian conditions present ‐ “Ideal”

80‐ 61 = high quality, some important pedestrian conditions present ‐ “Reasonable”

60‐ 41 = average quality, pedestrian conditions present but room for improvement ‐“Basic”

40‐ 21 = low quality, minimal pedestrian conditions – “Poor”

20 and below = poor quality, pedestrian conditions absent – “Not Suitable for Pedestrians”

We conducted detailed PEQI analyses for communities in and out of the zone to inform HIA recommendations.

These analyses are included in Appendix F, along with the PEQI Methods, and inform our recommendations for

prioritizing infrastructure improvements related to pedestrian safety and walking conditions in the next section of

this report.

Figure 21 depicts one of the maps of the PEQI findings for the streets and intersections surveyed in the congestion

pricing zone, followed by a summary of the PEQI findings for the SoMa neighborhood (labeled in the map).



Figure 21. Pedestrian Environmental Quality in the Congestion Pricing Zone

South of Market

In South of Market, 198 intersections and 646 street segments were assessed using the PEQI (see Figure 21). The

average PEQI intersection score was 28, which was the lowest score in the survey for area. Over 60 percent of the

intersections lack basic pedestrian amenities and engineering countermeasures and were ranked “not suitable for

pedestrians” by the PEQI (red intersections). Many of these intersections are junctions with short streets and alleys

where there are no crosswalks, signals, or signs warning pedestrians to look for cars; in some cases additional

pedestrian infrastructure might be appropriate and in other places not, based on factors including existing and

projected pedestrian and traffic volumes. The average PEQI street score for the streets was 47, also the lowest

scoring neighborhood in this survey. Twenty‐five percent of the streets were ranked “poor” (orange streets) or

“not suitable for pedestrians” (red streets). Many of the street segments that ranked “poor” are near the freeway,

with high traffic volumes and no traffic calming. Harrison is one street in particular that scored poorly throughout

South of Market.





Health Effects Characterization: Vehicle‐Pedestrian Injury Collisions

The following table summarizes our overall characterization of the health effects of the road pricing policy under

study on vehicle‐pedestrian injury collisions. Based on the available evidence: there is a high likelihood that the

proposed policy will impact on vehicle pedestrian‐injury collisions; severity of the effects are high and important

with respect to the health of San Franciscans; the magnitude of the predicted decrease in vehicle‐pedestrian injury

collisions is substantial relative to predicted increases under 2015 BAU; and the reduction of injuries particularly in

the specific census tracts in the cordon zone relative to the increases anticipated under 2015 BAU contributes to a

restorative equity effect as decreases are predicted in areas historically disproportionately adversely impacted by

pedestrian injury. Overall, given the assumptions and uncertainties discussed further below, we have a high‐

moderate degree of confidence in our characterization of these health effects.

Table 21. Road Pricing Policy Health Effects Characterization: Vehicle‐Pedestrian Injury Collisions Health Effects:

Characteristics


Overall

Assessment

Based on the following health effects characterization and

uncertainty factors regarding the magnitude of estimated effects,

what is the overall assessment of confidence in the health effects

estimate?

High‐Moderate


affect vehicle‐pedestrian injury collisions, irrespective of the


Very Likely/Certain: Adequate evidence exists for a

causal and generalizable effect based on the empirical

literature, evaluation studies of congestion pricing in

other countries.

Severity How important is the effect on vehicle‐pedestrian injury

collisions with regards to human function, well‐being or

longevity, considering the affected community's current ability to

manage the health effects of vehicle‐pedestrian injury collisions?

High: Reduces acute and permanent effects that are

potentially disabling or life‐threatening. 22% of all

traumatic injuries in San Francisco are caused by auto‐

versus‐pedestrian collisions.

Magnitude How much will vehicle‐pedestrian injury collisions change as a

result of the road pricing policy under study?Substantial (High‐Moderate Certainty):a Compared to

2015 BAU, 2015 with RP contributes to a decrease in

vehicle‐pedestrian injury collisions of 5% citywide, and of

9% in the cordon zone.

Distribution Will the effects be distributed equitably across populations? Will

the road pricing policy under study contribute to a reversal of

baseline inequities?

Restorative Equity Effects: Populations in the cordon

zone have been historically, disproportionately burdened

with vehicle‐pedestrian injury collisions in their

communities. The road pricing policy under study has the

potential to reduce some of this inequity.


Certainty in Health Effect Characterization

A summary of the principal uncertainties in our characterization of road pricing effects on vehicle‐pedestrian injury

collisions is provided in Table 22. The strengths and limitations of traffic volume and population forecasts which

are key inputs into the model and other analyses in this HIA are discussed at the beginning of Section V.

We have high‐moderate overall confidence in the forecasting model, which was developed using local injury data

and overall explains a substantial share (>70%) of variation in injury frequencies at the level of the census‐tract.90

The model further accounts for a well‐established non‐linear relationship between changes in pedestrian and

vehicle volumes and changes in injury frequency.91 Furthermore, predicted decreases in pedestrian injury

collisions under 2015 RP conditions are similar to the decreases seen with the introduction of congestion pricing in

London.92




Nevertheless, there are several important uncertainties to the precision of future estimates. Importantly, the

model was created based on observed counts of injuries. Observed counts of vehicle‐pedestrian injuries based

upon the collisions reported to and recorded by police generally underestimate the true burden of vehicle‐

pedestrian injuries. An analysis in San Francisco comparing data from 2000‐2001 police records with hospital data

from San Francisco General Hospital (SFGH, the City’s Level‐I Trauma Center which sees the majority of more

severely injured pedestrians) found that 22% of pedestrians injured and seen at SFGH were not reported in police

records.93 Undercounts of baseline injuries based on police record data thus will translate into undercounts of

predicted future injuries.

In addition, the model does not account for other pedestrian‐injury risk factors that may change with road pricing.

Specifically, data were not available to account for speed which is strong predictor of injury collision frequency.

The area‐level model also did not account intersection‐level street characteristics; however, we have no reliable

way of predicting how these countermeasures might change under future scenarios. Our analysis estimated

changes in all pedestrian‐collision injuries and not in injury severity. Speed is the most significant modifiable

predictor of pedestrian injury severity.94 The SFCTA has conducted analyses regarding estimated changes in

vehicle speeds with road pricing and found up to 25% increases in speeds on the most congested streets, from an

average of 10 to 12 mph. Conceivably, these increases in average travel speeds may have potential effects on

injury severity.

Table 22. Uncertainty Factors Regarding the Magnitude of Estimated Health Effects for Vehicle‐

Pedestrian Injury Collisions





Confidence Level

Exposure Assessment SFDPH model parameters (census‐tract level) include estimates

of the following conditions: traffic volume; proxies for

pedestrian activity (number of residents and employees,

neighborhood commercial zoning, land area, population living

in poverty, population aged 65 and older); proxies for traffic

speeds ( proportion of streets that are arterial streets without

transit). Attenuated per‐pedestrian risk with increasing

numbers of pedestrians is incorporated into the model

parameter for employees, which was fit using the natural log;

the traffic volume parameter is also a natural log consistent

with the empirical literature. Direct estimates of pedestrian

volumes, traffic speeds, and intersection‐level engineering

factors are not included as city‐wide data for those factors was

not available.

Moderate

Baseline Disease Prevalence Vehicle‐pedestrian injury collision data from the California

Highway Patrol's Statewide Integrated Traffic Records System

(SWITRS). Research in San Francisco has demonstrated that

SWITRS data underestimates total injury burden using hospital

data as a gold standard, with lower rates of under‐reporting for

more severe injuries (Sciortino et al. 2005). This uncertainty

results in an underestimate of the effect.

High‐Moderate


(ERF)

SFDPH model is specific to San Francisco, predicting over 70%

of the variation in pedestrian injury collisions in census tracts.

Model has not yet been validated over time. The model's area‐

level approach, indicative of small‐area hot spots, is

appropriate for area‐level policy assessment.

High‐Moderate



E. Vehicle-Cyclist Injury Collisions

Road Pricing Effects on Cyclist Safety

As stated in the previous section, traffic injuries and fatalities represent a significant and avoidable health burden,

with cyclists and pedestrians arguably the most vulnerable road users. In the U.S., cyclists experience a

disproportionate share of traffic injury and fatality risks compared to vehicle occupants on a per‐trip basis – with

estimated per‐trip fatality rates over two times that of motor vehicles (21.0 vs. 9.2 per 100 million person trips).95

Despite higher risk of injury per trip, cycling is a relatively low‐cost, physically active form of transportation that is

growing in popularity in the U.S. in recent years ‐ with an overall 64% increase in bike commuters in the U.S. from

1990‐2009 and with trends particularly notable in urban areas including San Francisco, Portland, Minneapolis,

Washington DC, Chicago, and New York City.96

The objective of road pricing is to reduce traffic congestion on local roadways during peak use periods, which also

correspond to times of high utilitarian cyclist volumes. Reductions in traffic volume will tend to reduce the

likelihood of vehicle‐cyclist injury collisions. However, greater use of cycles may increase the number of cyclists

exposed to traffic hazards. Where free flow conditions do not currently exist, reductions in traffic volume may also

increase traffic speed which is a risk factor for collision frequency and severity. Investments in road infrastructure

funded by road pricing could address gaps in cyclist safety engineering countermeasures, though could also

increase transit service on streets frequented by cyclists.

Figure 22 details a pathway through which road pricing policy impacts changes in vehicle trips and cyclist trips, and

therefore changes in traffic hazards and cyclists and ultimately vehicle‐cyclist injury collisions.

Figure 22. Pathway from Road Pricing Policy to Vehicle‐Cyclist Injury

Analysis of Vehicle‐ Cyclist Injury Collisions

We estimated expected changes in vehicle‐cyclist injury collisions using collision data from 2005 and changes in

motor vehicle and cyclist trips from the SFCTA’s model. We used a simple prediction model that assumes that

future injuries are a non‐linear function of changes in motor vehicle and cyclist trips (i.e., decreasing per‐cyclist risk

with increasing cyclists; increasing per‐cyclist risk with increasing vehicles; decreasing per‐motorist risk with

increasing vehicles). This approach is based on empirical studies of vehicle‐cyclist collisions and has been applied

to predict hypothetical scenarios in changes in collisions associated with changes in transportation mode share.97

Similar to the vehicle‐pedestrian injury models, changes in vehicle volumes have a stronger influence on vehicle‐

cyclist collisions than changes in cyclists volumes. A detailed description of methods is in Appendix G.





The map in Figure 23 depicts intersection‐level

vehicle‐cyclist injury collisions under actual 2005

conditions. The concentration of collisions in the

potential congestion pricing zone is evident – with

approximately 50% of vehicle‐cyclist collisions

occurring in the congestion pricing zone.

Cyclist injury and collisions – including vehicle‐

cyclist injury collisions – have been increasing in

San Francisco in recent years. Based on data on

police reported injuries from SWITRS for 2006 –

2009, there were 273 injuries to cyclists in

collisions with motor vehicles reported by the

police in 2006, climbing to 424 in 2009 – a 55%

increase. Because of under‐reporting of cyclist

injuries to police, these statistics are likely to

represent an undercount of the true number of

vehicle‐cyclist injuries in San Francisco. These

increases parallel those in cycling in San Francisco. For example, the San Francisco Municipal Transportation

Agency’s annual citywide bicycle count found that counts of observed cyclists increased by 53% during those same

years, from 5,500 in 2006 to 8,441 in 2009.98

Figure 23. Vehicle‐Cyclist Injury Collisions in 2005:

San Francisco, California

The following table summarizes the predicted changes in vehicle‐cyclist injury collisions citywide as well as in the

cordon zone.

Table 23. Annual Estimated Vehicle‐Cyclist Injury Collisions in 2005, 2015 Business As Usual, and 2015


Health Impacts

(Annual Estimates)

2005 2015 BAU 2015 RP Change:

2005 →

2015 BAU

Change:

2005 →

2015 RP

Change: 2015

BAU → 2015

RPVehicle‐Cyclist Injury Collisions (N, Total)

Citywide 270 295 290 25 20 ‐5Congestion Pricing Zone 135 155 150 20 15 ‐5

2005: In 2005 conditions, the annual number of police‐reported vehicle‐cyclist injury collisions in San Francisco is

approximately 270. Approximately half of San Francisco’s vehicle‐cyclist injury collisions occur in the cordon zone

that is under study. The high‐density of vehicle‐cyclist injury in the proposed congestion pricing zone is explained

largely by higher volumes of traffic and cyclists in this zone.

2005 → 2015 BAU: In 2015 BAU, based on the model, we estimate vehicle‐cyclist injury collisions will increase 9%

citywide to 295 total vehicle‐cyclist injury collisions, with a 15% increase in the potential congestion pricing zone

accounting for 80% of the citywide increase. Similar to the baseline conditions, approximately 50% of total injuries

are expected to occur in the congestion pricing zone. Increases in the number of injury collisions are related to

predicted increases in traffic and cycling to and from San Francisco’s downtown core.

2015 BAU → 2015 RP: We estimate that vehicle‐cyclist injury collisions could also increase relative to baseline

conditions under a future road pricing scenario; however the increase will be smaller than in the business as usual



scenario. Decreases in vehicle traffic volume citywide and in the cordon zone explain the moderation of the

expected injuries.

Because the congestion pricing scenario largely reduces traffic volumes during peak commute periods, as with

pedestrian injury collisions we also assessed the time of day of vehicle‐cyclist injury collisions in San Francisco. We

found that approximately 30% of vehicle‐cyclist injury collisions occur during the AM and PM peak commute

periods, supportive of our finding that changes in traffic volume during those peak commute periods could have a

protective effect on vehicle‐cyclist injury collisions.

Health Effects Characterization: Vehicle‐Cyclist Injury Collisions


vehicle‐cyclist injury collisions. Based on the available evidence: there is a high likelihood that the proposed policy

will impact on vehicle cyclist‐injury collisions; severity of the effects are high and important with respect to the

health of San Franciscans; the magnitude of the change in vehicle‐cyclist injury collisions among future scenarios is

substantial with 2015 RP moderating the predicted increase of injuries expected under 2015 BAU. The largest

moderating effects on cyclist injuries occurs within the cordon zone contributing to a restorative equity effect as

decreases are predicted in areas historically disproportionately adversely impacted by cyclist injury. Overall, given

the assumptions and uncertainties discussed further below, we have a moderate‐low degree of confidence in our

characterization of these health effects. The estimated increase in vehicle‐cyclist collisions in both future scenarios

makes a case for investments in additional protective engineering countermeasures for cyclists in San Francisco.

Section VI provides several recommendations for measures to improve cyclist safety.

Table 24. Road Pricing Policy Health Effects Characterization: Vehicle‐Cyclist Injury Collisions Health Effects:

Characteristics


Overall

Assessment

Based on the following health effects characterization and

uncertainty factors regarding the magnitude of estimated effects,

what is the overall assessment of confidence in the health effects

estimate?

Moderate‐Low


affect vehicle‐cyclist injury collisions, irrespective of the


Very Likely/Certain: Adequate evidence exists for a

causal and generalizable effect based on the empirical

literature and evaluations from implementation of

congestion pricing in other countries.

Severity How important is the effect on vehicle‐cyclist injury collisions

with regards to human function, well‐being or longevity,

considering the affected community's current ability to manage

the health effects of vehicle‐cyclist injury collisions?

High: Reduces acute and permanent effects that are

potentially disabling or life‐threatening.

Magnitude How much will vehicle‐cyclist injury collisions change as a result

of the road pricing policy under study?Substantial (Moderate‐Low Certainty):a Compared to

2015 BAU, 2015 with RP contributes to a decrease in

vehicle‐cyclist injury collisions of 3% citywide and in the

cordon zone.

Distribution Will the effects be distributed equitably across populations? Will

the road pricing policy under study contribute to a reversal of

baseline inequities?

Restorative Equity Effects: The cordon zone area has a

historically high burden of vehicle‐cyclist injury collisions.

The road pricing policy under study has the potential to

reduce some of this inequity.






Certainty in Health Effect Characterization

A summary of the principal uncertainties in our characterization of road pricing effects on vehicle‐cyclist injury

collisions is provided in Table 25. The strengths and limitations of traffic volume and population forecasts which

are key inputs into the model and other analyses in this HIA are discussed at the beginning of Section V. Overall,

we have limited to moderate confidence in the precision of estimates of future cyclist injuries predicted by the

model. For several important reasons, discussed below, our modeling approach may be over or under‐estimating

future bicycle injury frequencies.

First, the model could be over‐estimating injuries if future improvements in road infrastructure reduced hazards

for bicyclists. Our estimation approach based on vehicle and cycle volumes alone does not account for bicycle

infrastructure or safety countermeasures that might change in the future under 2015 BAU or road pricing

conditions. Bicycle infrastructure is expected to improve citywide over the next years due to programmed

improvements. However, at present, there is limited evidence on the protective effects of bicycle lanes on urban

streets so the ultimate safety effects of this infrastructure cannot be predicted.99

Importantly, future estimates of cyclist injuries are a function of baseline frequency of injuries. Baseline counts of

vehicle‐cyclist injuries based upon the collisions reported to and recorded by police generally underestimate the

true burden of vehicle‐cyclist injuries. Undercounts of baseline injuries thus will translate into undercounts of

future injuries.

There are several other reasons that our modeling approach could underestimate the future change in cyclist

injuries. Most important, the SFCTA model could be underestimating the future changes in cycling. Notably, the

actual observed increase in vehicle‐cyclist collisions from 2006 to 2009 (55% change) is over three times the

increase predicted by our model from 2005 to 2015. A likely explanation for this increase is an increase in cycling

not accounted for by the model. Other explanations for this increase in reported cyclist injuries include an

increase in the proportion of collisions being reported to police and changes in driver or cyclist safety behaviors.

Our predictions clearly underestimate the full burden of cyclist injuries in baseline and future scenarios as we have

only accounted for and modeled changes in injuries associated with vehicle‐cyclist injury collisions, excluding

injuries to cyclists that do not involve an occupied motor vehicle. Causes of cyclist only injuries may include cyclists

swerving to avoid a collision and hazardous roadway conditions (e.g. potholes). A recent analysis by the San

Francisco Injury Center of patients seen by the San Francisco General Hospital (SFGH) for trauma sustained while

cycling found that approximately 25% of those patients were injured in a “cyclist only” crash that did not involve

contact with a motor vehicle. These “cyclist‐only” crashes were notably under‐reported in the police‐reported

SWITRS data historically relied upon by the City to monitor collisions over time. Specifically, 22% of SFGH patient

records did not matched with a SWITRS record; of those SFGH records, 58.5% were cyclists injured in a crash with a

motor vehicle and 41.5% were of cyclist‐only injuries. Notably, 42% of cyclist injuries reported by SWITRS did not

match with a SFGH record; of those SWITRS records, 91% were injured in a crash with a motor vehicle and 9% were

cyclist‐only injuries.100

Our estimation approach predicts total vehicle‐cyclist injury collisions and does not account for any effects on collision severity. Changes in travel speeds associated with infrastructure changes or changes in vehicle volume may increase or decrease speed, thereby affecting injury severity. As summarized in the previous section, the SFCTA has conducted analyses regarding estimated changes in vehicle speeds with road pricing and found up to 25% increases in speeds on the most congested streets, from an average of 10 to 12 mph. Conceivably, these increases in average travel speeds may have potential effects on injury severity.



Table 25. Magnitude of the Estimated Health Effects for Vehicle‐Cyclist Injury Collisions: Uncertainty

Factors


Confidence Level

Exposure Assessment A simple forecasting approach based on estimates of motor

vehicle trips and cycling trips in 2005 and 2015 conditions.

Intersection‐level engineering factors, recreational cycling trips

not considered in the analysis.

Moderate‐Low

Baseline Disease Prevalence Vehicle‐cyclist injury collision data from the California Highway

Patrol's Statewide Integrated Traffic Records System (SWITRS).

Likely an underestimate of total injury collisions.

Moderate


(ERF)

Model parameters informed by summary of international

findings regarding the relationship between changes in motor

vehicle and cyclist trips and vehicle‐cyclist injury. There is

consistency in the findings regarding a non‐linear relationship

though variance in parameter values (accounted for with

sensitivity analyses). Variance in model parameters potentially

due to regional differences in environmental and/or socio‐

demographic factors not accounted for in the model.

Moderate‐Low





F. Health Inequities

Road Pricing and Health Inequities

Health inequities are defined as systematic disparities in health status or in the major social determinants of health

between groups with different social advantage/disadvantage (e.g., wealth, power, prestige).101 Transportation

policies have several potential effects on determinants of health including the quality, accessibility, and cost of

transportation, transportation hazards, and exposure to environmental hazards, neighborhood livability, and

opportunities for physical activity. Policy effects in any of these domains may favor or be adverse to specific

populations experiencing or vulnerable to health inequities.

The SFCTA preliminary studies considered potential equity impacts of the fee on low‐income drivers, as detailed in

the Scoping section (III.D.) of this report. These studies concluded that a congestion pricing program would “not

have an undue impact on low‐income individuals.”102 This assessment was based on their study recommendations

that the policy include a 50% discount for low‐income travelers; that approximately 5% of peak‐period travelers to

downtown San Francisco are drivers with annual household incomes < $50,000; and that many low‐income

travelers in the Bay Area already rely on public transit and would thus benefit from faster, more reliable transit

travel times resulting from pricing fee investments. The SFCTA further found that support for studying a San

Francisco congestion pricing program is highest among low‐income Bay Area residents.

This HIA analyzed the health effects of road pricing with respect to spatial equity, assessing effects on air and noise

pollution, physical activity, and transportation‐related injuries city‐wide and within the congestion pricing area

under study, as detailed in each of the distinct impacts analyses subsections. This section contributes an additional

analysis of whether these effects are equitably distributed under the road pricing scenario among populations

defined by household income or by age. There are several reasons to consider the particular effects on children

and youth, seniors, and low‐income populations. Children and adolescents are particularly vulnerable to traffic‐

related health exposures due to their ongoing growth and development – with lungs still maturing and particularly

susceptible to air pollution, with brain development underway and adversely impacted by traffic‐related noise and

disturbance to sleep and the ability to concentrate, and with cognitive functions unable to fully comprehend and

respond to potentially fatal traffic hazards. Seniors are often more dependent on non‐motorized transportation

for their daily needs, and are more likely to die in vehicle‐pedestrian injury collisions – over six times more likely

based on recent analyses in San Francisco. Low income populations have been historically, disproportionately

exposed to traffic‐related hazards and their adverse health impacts. Additionally, these communities typically are

confronted with other social and environmental stressors which cumulatively contribute to disproportionate

adverse health effects, including: poor housing quality; stressful work environments; and limited economic

resources to meet basic needs for housing, food, transportation, and health care.



Equity of Traffic‐Related Health Hazardous Environmental Exposures for children and youth, seniors,

and low‐income populations

We used traffic density as a general proxy for adverse environmental exposures and hazards of traffic. The

intensity of vehicle air pollution emissions, traffic noise, and safety hazards to non‐motorized users are all

generally proportional to the density and proximity of vehicles in an area. We created a summary measure of

traffic density at the small‐area transportation analysis zone (TAZ) level to assess area‐level traffic exposure in

2005, 2015 BAU, and 2015 RP conditions using ArcGIS mapping tools (see methodological footnote, below).b

Traffic density maps were presented earlier in this report (Figure 9). Table 26 summarizes the total population

exposure in the three study scenarios for three traffic density categories, with categories consistent for each

scenario and cut‐points created based on the TAZ‐level distribution of traffic density in 2005 conditions.

Table 26. The Distribution of Traffic Density in the General Population: 2005, 2015 Business As Usual,

2015 with Road Pricing

2005 2015 BAU 2015 RP

Traffic density

Population, N

= 796,000

Population, N

= 824,000

Population, N

= 824,000

At or below average traffic density 67% 65% 66%

1 ‐ 2x average traffic density 24% 25% 26%

>2x average traffic density 9% 10% 9%

Overall, there is a moderate increase in the proportion of the population exposed to higher traffic density from

2005 to 2015 BAU conditions. For 2015 RP relative to 2005, there is a smaller population increase in traffic

exposure. These findings are consistent with the analysis of air pollutant and noise impacts.

Traffic Density Distributions by Age

The following table summarizes traffic density by age in 2005, 2015 BAU, and 2015 RP conditions.

Table 27. Traffic Density by Age: 2005, 2015 Business As Usual, 2015 with Road Pricing

Youth:

Ages 0‐19

Adults:

Ages 20‐64

Seniors:

Ages 65+

Youth:

Ages 0‐19

Adults:

Ages 20‐64

Seniors:

Ages 65+

Youth:

Ages 0‐19

Adults:

Ages 20‐64

Seniors:

Ages 65+

N= 126,000 N= 563,000 N= 107,000 N= 127,000 N= 575,000 N= 122,000 N= 127,000 N= 575,000 N= 122,000

At or below average

traffic density 71% 66% 66% 70% 64% 64% 71% 65% 66%1 ‐ 2x average traffic

density 19% 25% 25% 19% 26% 24% 20% 27% 25%>2x average traffic

density 9% 9% 9% 11% 10% 11% 9% 8% 9%

2005 2015 BAU 2015 RP

Under 2005 conditions, youth aged 0‐19 have a higher proportion of their population (71%) represented in areas

with less than average traffic density, relative to adults and seniors. However, youth are proportionally

b We first converted street segment traffic volumes to a density metric using a 100‐meter grid size and the ArcGIS kernel

density method. We next converted the ArcGIS raster data to point data, and assigned the density values to their corresponding

TAZ using a spatial join, aggregated the values, and finally normalized the data by the TAZ area. We used 2005 baseline traffic

density conditions by TAZ to create the average traffic density, 1‐2x average traffic density, and >2x average traffic density

categories used in the analysis. We selected these categories based on the distribution of traffic density in San Francisco.





represented in the highest traffic density areas. Traffic density exposure for the elderly is similar to that as that of

the overall adult population under 2005 conditions. Figure 6 in Section IV illustrates youth and senior population

densities in San Francisco areas under 2005 conditions; as evidenced in the maps, there are higher concentrations

of youth in western San Francisco where there are lower concentrations of traffic density.

The relative differences in exposure by age remain largely unchanged under both future 2015 scenarios (BAU and

RP) – with a higher proportion of youth in the less than average exposure category relative to adults and seniors in

all three scenarios, but similar proportions in the highest exposure category for all three age groups. In fact, the

2015 RP scenario provides a moderately protective equity effect by reducing exposure across all age groups

relative to 2015 BAU.

Traffic Density Distributions by Average Area Household Income

The following table summarizes traffic density by average area household income in 2005, 2015 BAU, and 2015 RP

conditions.

Table 28. Traffic Density by Income: 2005, 2015 Business As Usual, 2015 with Road Pricing

Avg. HH

Income:

<$35,000

Avg. HH

Income: $35‐

70,000

Avg. HH

Income:

>$70,000

Avg. HH

Income:

<$35,000

Avg. HH

Income: $35‐

70,000

Avg. HH

Income:

>$70,000

Avg. HH

Income:

<$35,000

Avg. HH

Income: $35‐

70,000

Avg. HH

Income:

>$70,000

N= 98,000 N= 496,000 N= 202,000 N= 88,000 N= 487,000 N= 249,000 N= 88,000 N= 487,000 N= 249,000

At or below average

traffic density 31% 70% 77% 25% 69% 71% 25% 70% 72%1 ‐ 2x average traffic

density 64% 20% 15% 67% 20% 20% 68% 21% 20%>2x average traffic

density 5% 9% 9% 8% 11% 9% 7% 9% 8%

2005 2015 BAU 2015 RP

Differences in exposure to traffic density for each income group are notable under baseline conditions. Compared

to the highest income households (with average household income >$70,000) fewer low‐income households

(<$35,000) live in areas with less than average traffic density (77% vs 31%). Although low‐income households are in

areas with lower levels of the highest average traffic density (5% vs. 9%), overall a higher percentage of the low‐

income households are in areas with average or above average traffic density compared to higher income

households (69% vs. 24%).

Figure 5 (Section IV) illustrates average household income by TAZ in San Francisco under 2005 conditions. Visually

comparing that map to the traffic density maps in Figure 9 reveals that area‐level concentrated poverty (as

measured by average household income <$35,000) is associated with higher area traffic density. This may be the

result of past transportation and land use planning and policy decisions which increased traffic on the streets of

lower income communities.

Future conditions under 2015 BAU estimate traffic density increases for all income categories. Among low income

households, fewer are predicted to live in lower traffic areas (25% vs. 31%) and more could live in higher traffic

areas (8% vs. 5%). 2015 with road pricing contributes to relatively moderate decreases in traffic density exposure

citywide. There is still a decrease in the proportion of low income households experiencing lower traffic density

under 2015 RP; however, the increase in the proportion of low‐income households in high traffic density areas (7%

vs. 5%) is smaller than under 2015 BAU conditions (8% vs. 5%).




Of note, areas where the lowest income communities experience higher concentrations of traffic are also areas

with the lowest prevalence of motor vehicle ownership and, as detailed in the Active Transportation analysis –

areas where more residents walk (or rely on transit) for transportation. As described in the vehicle‐pedestrian

injury collision and vehicle‐cyclist injury collision analysis sections, these communities are currently

disproportionately burdened with injuries to vulnerable road users.

Overall, under 2005 conditions low‐income households are disproportionately burdened with environmental

exposures related to higher traffic densities ‐ including air pollution, noise, and pedestrian safety hazards. Based

on this analysis, 2015 with road pricing is expected to mitigate some increased exposure to traffic density

predicted under 2015 BAU. As discussed in the rest of this section (V.), the road pricing policy under study has a

very moderate beneficial effect in reducing exposure across all populations.




G. Costs

Economic Valuation of Transportation‐Attributable Health Outcomes

This section provides our analysis of the economic value of the health effects analyzed in prior sections. We

assessed the economic cost or benefit of transportation‐related health effects only for those health effects that

were quantified and that had accepted monetary valuation methods. The approach used is generally consistent

with the practices used in regulatory impact assessment and cost benefit analysis by the U.S. Environmental

Protection Agency (USEPA) and other regulatory impact assessments.103

The economic value of health effects can include direct health care costs as well indirect costs such as lost wages

for individuals, lost productivity for employers, pain and suffering for individuals, relatives, and family members.

Furthermore, changes in the health of one individual can also affect the health and well‐being of another, resulting

in additional economic costs. There are a range of methods available to assess the economic value of health

effects. Methods differ in their ability to capture the range of costs of health outcomes, including costs that do no

have economic markets. For example, some of the earliest methods to value life in assessments of pre‐mature

mortality were based only on assessments of effects on lost economic productivity or income and gave no value to

“intangibles” such as grief. In contrast, hedonic economic methods attempt to capture such intangible and non‐

market effects by using observations of economic behaviors (e.g. trade‐offs among wages and hazardous

employment, the cost of the purchase of consumer safety products) to reveal what value individuals give to

protecting life. The hedonic method makes important assumptions about individual knowledge and employment

choices. The contingent valuation method assumes a hypothetical market and asks individuals to state their

willingness to pay for a change in the risk of mortality, typically via a survey. Although each of the above methods

has limitations, the contingent valuation method appears to have become the preferred approach to value life in

the context of regulatory analysis.104

A summary of economic values of health effects, valuation methods, and sources used to value health effects is

described below and provided in Table 29.



Table 29. Economic Valuation of Health Effects: Input and Methods

Health

Outcome

Economic

Valuation

Valuation

Year

Adjusted

2010 Year

Value

Economic Valuation

Method

Agency Source

Adverse Effects

7,400,000$ 2006 8,004,040$ Meta‐analysis of wage‐

differential and stated

preference studies

USEPA

N/A N/A N/A Estimates for U.S.

populations not available

N/A

262,600$ 2000 332,529$ Sum of Lifetime Medical

Cost, Lost Productivity, and

Present value of Earnings;

age 65 female and 70 male

respectively; average of

male and female

Harvard Center

for Risk Analysis



preference studies

USEPA

Minor 14,800$ 2006 16,008$

Moderate 114,700$ 2006 124,063$

Serious 425,500$ 2006 460,232$

Severe 1,387,500$ 2006 1,500,757$

Critical 5,642,500$ 2006 6,103,080$



preference studies

USEPA

Mortality averted due to physical activity via walking and biking for active transportation

Beneficial Effects

Avoidable mortality attributable to traffic‐related fine particulate matter (PM 2.5)

Avoidable mortality attributable to vehicle‐pedestrian and vehicle‐cyclist collisions

Avoidable non‐lethal injuries by severity attributable to vehicle‐pedestrian and vehicle‐cyclist

Estimated as a fraction of

the VSL based on the

severity of the injury

USDOT

Annoyance due to traffic‐related noise

Myocardial infarction cases attributable to traffic‐related noise

For all impacts on life expectancy , including traffic fatalities, pre‐mature mortality from particulate matter

exposure, and reduced mortality from improved physical activity, we used the USEPA default central “value of

statistical life” (VSL) of $7.4 million (2006$). 105 Notably, the VSL used by the USEPA is different than the VSL values

used by some other Federal agencies including the USDOT.





For ischemic heart disease events we used a valuation combining three types of costs of non‐fatal myocardial

infarctions: lifetime medical cost, lost productivity, and lost wages.106 This valuation approach does not capture

the value of averting the illness or other intangibles. We assumed that none of the ischemic heart disease events

would be fatal.

To assess the value of non‐fatal traffic injuries, we used a method following on US DOT guidelines for economic

valuation107 that scales the value of injuries as a fraction of the VSL based on how the injury affects of the quality

of life. Injuries are categorized along a Maximum Abbreviated Injury Scale (MAIS), into levels ranging from MAIS 1

(minor) to MAIS 5 (critical). Table 30 provides the parameters used to multiply by the current value of preventing a

fatality to obtain the value of preventing injuries of the relevant types.

Table 30. U.S. Department of Transportation Guideline Values for Lethal and Non‐Lethal Injuries

MAIS Level Severity Fraction of VSL

MAIS 1 Minor 0.0020

MAIS 2 Moderate 0.0155

MAIS 3 Serious 0.0575

MAIS 4 Severe 0.1875

MAIS 5 Critical 0.7625

MAIS 6 Fatal 1.0000

We estimated the annual economic value of transportation‐attributable health effects under each of the three

transportation scenarios—2005 baseline, 2015 BAU, and 2015 RP using the outputs of the quantitative health

effects analysis described in this report (see Executive Summary table). We assumed that the distribution of the

severity of vehicle‐pedestrian and vehicle‐cyclist injury collisions would remain similar to the distribution under

2005 baseline conditions. We used injury severity assigned in SWITRS to assign the above MAIS codes as follows,

using the proportion of motor vehicle‐pedestrian and motor vehicle‐cyclist injuries seen by San Francisco General

Hospital, San Francisco’s only trauma center, to inform the proportion of injuries that were critical (MAIS 5) and

severe (MAIS 4). We conducted analysis for the scenario year only and thus did not discount future effects.

Table 31 provides estimates of the economic value of costs of motor vehicle use and the health benefits of non‐

motorized transportation. The total aggregate annual economic value of transportation‐attributable adverse

health effects was lowest in 2015 RP ‐ $4 million lower than costs estimated for 2005 and $53 million less than

costs estimated for 2015 BAU. Vehicle‐pedestrian and vehicle‐cyclist injury collisions contributed the most to the

economic value of adverse health effects under all scenarios – accounting for just over half of the estimated

health‐associated costs. Premature mortality attributable to air pollution was the next most important

contributor, accounting for approximately 45% of costs. Approximately 2% of the enumerated costs were

associated with heart attacks attributable to traffic‐related noise.



Table 31. Annual Economic Values of Transportation‐Attributable Health Effects under 2005, 2015

BAU, and 2015 RP scenarios: San Francisco, California

2005 2015

BAU

2015 RP 2005 2015 BAU 2015 RP

Adverse Effects Value: Total 1,124$ 1,173$ 1,120$

Avoidable mortality attributable to traffic‐related fine particulate matter (PM 2.5)

$8,004,000 65 66 63 520$ 528$ 504$

Annoyance due to traffic‐related noise

N/A N/A N/A N/A N/A N/A N/A

Myocardial infarction cases attributable to traffic‐related noise

$332,529 31 34 34 10$ 11$ 11$

Avoidable injuries by severity attributable to vehicle‐pedestrian and vehicle‐cyclist collisions

Minor $16,008 575 615 587 9$ 10$ 9$

Moderate $124,062 196 210 201 24$ 26$ 25$

Serious $460,232 196 210 201 90$ 97$ 93$

Severe $1,500,757 56 60 57 84$ 90$ 86$

Critical $6,103,000 37 40 38 228$ 243$ 232$

Fatal $8,004,000 20 21 20 158$ 168$ 160$

Beneficial Effects Value: Total 1,225$ 1,305$ 1,337$

Mortality averted attributable to biking for active transportation

8,004,000$ 23 25 26 184$ 200$ 208$

Mortality averted attributable to walking for active transportation

8,004,000$ 130 138 141 1,041$ 1,105$ 1,129$

Green: estimates of beneficial cost savings from mortality averted from active transportation via

walking and biking.

Health

Outcome

Economic Value

of Health

Outcome ($2010)

Cases / Events Estimated Economic Value

(Millions of $)

Estimated beneficial cost savings from mortality averted from active transportation via walking and biking were

slightly greater than the estimated adverse health costs in all scenarios, with an estimated $1.34 billion in 2015 RP,

$1.31 billion in 2015 BAU, and $1.23 billion in 2005. Beneficial cost savings are greatest under the 2015 RP

scenario.

Limitations

This economic valuation of transportation‐related health effects has several important limitations.

First, this economic analysis did not count all of the health and welfare outcomes attributable to

transportation. We did not quantify each health effect identified, (for example, asthma related to vehicle

emissions or cognitive impairment due to noise) and we did not quantify physical activity effects on

chronic disease morbidity.

Second, as discussed above, we could not identify suitable economic values for each health effect

quantified. Specifically, we did not identify published U.S. values for the noise‐related health effect of

subjective annoyance. While valuation methods exist for noise and health outcomes based on studies in

European countries, we could not find examples of economic values of noise‐related health effects in the

U.S. literature or examples of noise‐related health costs in U.S. Cost‐benefit analysis. We did not apply





the European values to the U.S. population in the interest of consistency in valuations used for our

economic analysis.

Third, as with all economic valuations of health and welfare outcomes, this analysis has several general

conceptual limitations. The reasonability of placing a monetary price on health or life and assessing the

value of health to future generations is a topic of ongoing debate.108

Finally, the economic costs and benefits should not be construed to represent a complete cost‐benefit

analysis of the proposal. For example, we did not consider the include the direct costs of transportation

of the value of time spent or saved as a result of the pricing program.




VI. Recommendations

Recommendations for Maximizing the Health Benefits of Road Pricing Policy

This section provides recommendations for investments and strategies to support the success of road pricing and

maximize its health benefits. Identifying feasible and effective recommendations for policy or decision alternatives

and mitigations is a key part of the HIA process. In general, road pricing as designed has moderate health and

equity benefits relative to business as usual scenarios. As the negative health consequences of road pricing are

few, the following recommendations are provided for investments and strategies that would complement road

pricing as a health improvement strategy. Further, the HIA findings provide direction with respect to targeting

policies and mitigations to specific locations to address potential adverse impacts and enhance health benefits.

One of the specific benefits of road pricing is that it generates revenue that may be used to maintain and improve

transportation infrastructure. An objective of road pricing is to reduce the negative impacts of private motor

vehicles on the transportation system, it is therefore appropriate that revenues from road pricing be used to

improve the quality and safety of transportation for transit and non‐motorized transportation users. Many such

investments can be leveraged for improving health.

We propose the following recommendations based on reviews of the existing interdisciplinary literature and

consultation with experts in transportation, economics, and environmental planning – including those who are

consultants on our study team. SFDPH and other members of the research team have substantial experience in

formulating recommendations for transportation and land use policy base on experience with HIA and through the

development and application of the healthy development measurement tool (HDMT; www.thehdmt.org). In

formulating these recommendations, it was critical to include consideration of investments with particular benefits

for transit‐dependent populations and neighborhoods that have been historically burdened by adverse impacts of

the transportation system, because this new revenue can redress those injustices. Our recommendations are not

exhaustive, and further study could likely identify additional best practices.

A number of these recommendations are already under consideration by the SFCTA as programmatic

improvements or mitigations to potentially enhance travel options or mitigate/minimize traffic and related

environmental impacts as part of a comprehensive program and investment package. Additionally, these

improvements – and their potential impact – have not been identified at this stage of the SFCTA’s study of the

potential congestion pricing policy, nor were they included in the travel model outputs that informed this HIA. The

SFCTA travel model does not currently have the capacity to account fully for all expected investments. Similarly,

this HIA has not prospectively modeled any physical environmental changes that may result from road pricing

revenue investments. These investments could have significant independent health effects that could be assessed

in a separate, future analysis.

A number of these recommendations are consistent with and/or would support other ongoing, inter‐agency

efforts currently underway in the City and County of San Francisco. SFDPH is engaged with sister agencies in work

to assess and address transportation system‐related health effects, including the Citywide Pedestrian Safety Task

Force and the Community Risk Reduction Plan to improve air quality. Partnerships among City agencies, including

SFDPH, would be used to help finalize any package of improvements to accompany the implementation of pricing.




Air Quality

We expect that the road pricing policy may moderate air pollution increase in areas with expected growth in

vehicle air pollution emissions. There are not cumulative negative air quality consequences requiring mitigation,

however, augmenting the fee or extending the fee to other times may amplify these benefits ‐ specifically:

Consider augmenting the fee on days with poor air quality (e.g. BAAQMD spare the air days).

Consider augmenting the fee for vehicles with relatively greater pollutant emissions per passenger‐mile.

Noise

As discussed in the impact analysis section, we expect noise will modestly increase in all scenarios and this increase

is likely to be greatest along transit corridors. Following are recommendations to moderate these increases:

Consider applying the fees to motorcycles.

Consider augmenting the fee for the noisiest vehicles, including large commercial freight trucks.

Consider road pricing investments in quieter transit vehicle or road surface technologies.

Consider utilizing the fee to provide subsidies for residential acoustical protection along the noisiest

thoroughfares.

Active Transportation

Pricing policies will result in shifts to transit and non‐motorized transport, increasing physical activity. Augmenting

the fee or extending the fee to other times may amplify these benefits. Improvement strategies for pedestrian

safety and cyclist safety, cited below, would likely enhance the benefits of the policy on physical activity. Specific

recommendations for physical activity are:

Consider using fees to augment cyclist – transit facilities/infrastructure to improve active transportation

connections, particularly for the western part of the city with the lowest active transportation rates and

the farthest distance to travel from downtown (e.g., secure bike parking near transit).109

Consider using fees in support of employer‐based transportation demand management programs to

incentivize commuting to work via walking, biking and public transit and decrease driving.110

Adding street trees, public seating, pedestrian scale lighting, and enhancing public art on the sidewalks.111




Pedestrian Safety

Pricing reduces road volume and consequentially one of the most important determinants of injury frequency.

Pricing policies also will result in shifts to transit and non‐motorized transport, increasing the number of

pedestrians exposed to hazards from motorized vehicles. Reducing congestion may also increase vehicle speeds

on local roadways. Policies and investments could include enforcement and engineering strategies to mitigate

these effects and improve safety for non‐motorized users.

Specific recommendations include:

Consider investing road pricing revenues in traffic safety enforcement within and at the edge of the

congestion pricing zone, combined with media messages targeted at reducing hazardous driving

behaviors.

Consider investing road pricing revenues to implement pedestrian safety engineering strategies along

identified high‐risk pedestrian corridors within and at the boundary of the congestion pricing zone – with

particular attention to streets near senior centers, schools, health centers, and other key community

facilities that attract vulnerable users.

Based on the PEQI data analysis, South of Market is most in need of preventive and anticipatory

environmental design for pedestrians. Specific changes in SoMa and other high‐injury areas could

include: 112

Reducing the number of lanes or narrowing the roadway on multilane roads

Adding traffic calming measures along high‐risk corridors and streets including: lane narrowing, signal

timing, dynamic speed display signs, speed radar trailers, and street trees

Adding mid‐block, signalized crossings on long blocks

Extending signalized crossing times for pedestrians

Adding curb‐extensions and median islands

Installing high visibility crosswalks and advance yield lines

Adding pedestrian‐scaled lighting

Reducing and restricting curb cuts for motor vehicles

Providing beacons or hybrid signals at key crossing locations

Relocating bus stops to safe crossing locations

Providing enhancements to traffic signals, such as leading pedestrian intervals, scramble phases,

protected left turns, or prohibited right turns on red

Further target pedestrian safety engineering countermeasures and environmental improvements to roadways,

particularly arterials, outside the congestion pricing zone with existing high numbers of pedestrian injuries and

fatalities (see Figure 20) anticipated to experience additional traffic under 2015 BAU and/or 2015 RP conditions.

Specific improvements to consider include:113

Systematically installing pedestrian signals, crosswalks (high visibility crosswalks when warranted )

and bulb outs at key intersections along neighborhood commercial corridors

Installing evenly‐lit pedestrian‐scaled lighting that makes pedestrians visible

Reducing and restricting curb cuts for motor vehicles

Adding directional curb ramps to all intersections




Cyclist Safety

Pricing reduces road volumes and consequentially one of the most important determinants of injury frequency. As

stated above, pricing policies also will results in shifts to transit and non‐motorized transport, increasing the

number of cyclists exposed to hazards from motorized vehicles. Reducing congestion may also increase vehicle

speeds on local roadways. Policies and investments could include enforcement and engineering strategies to

mitigate these effects and improve safety for non‐motorized users. In existing conditions, cycling and cyclist

injuries have been increasing. Revenue investments in resurfacing streets commonly used by cyclists help to

reduce environmental hazards (e.g., uneven surfaces). Specific recommendations for cyclist safety include:114

Consider investing in engineering measures that reduce injury risks to cyclists from motor vehicle

collisions – particularly on busy arterials with higher cyclist volumes or designated bike routes –

including the traffic calming measures cited above and bike boxes at intersections.

Consider investing in separated bikeways via parking or other physical buffers, particularly on busy

arterials, truck routes, or transit routes with higher cyclist volumes

Consider targeted enforcement efforts to keep bike facilities clear of motor vehicles and improve

motorist and cyclist behavior/compliance

Monitoring

Monitoring the implementation of the policy, should it be implemented, is an important strategy for assessing its

impacts and addressing any unanticipated effects related to health and equity. Specific recommendations for

monitoring include:

Monitoring indicators of the distribution of health, geographic, and other equity‐related effects,

including: local traffic volumes and speeds, pedestrian and cyclist injuries, noise complaints, cut‐

through traffic complaints, noise and air pollutant levels, pedestrian and cyclist volumes, use of policy

discounts by low‐income and other identified populations, and other indicators as appropriate.




References 1 WHO (World Health Organization). 2011. Health in the Green Economy: Transport Sector. Geneva, Switzerland: World Health Organization. Available at: www.who.int/hia/hgebrief_transp.pdf.

2 Dora C, Phillips M. 2000. Transport, environment and health. Copenhagen, Denmark: World Health Organization Regional Office for Europe.

Ewing R, Frank L, Kreutzer R. 2006. Understanding the Relationship between Public Health and the Built Environment: A Report to the LEED‐ND Core Committee. Available at: www.usgbc.org/ShowFile.aspx?DocumentID=1736.

HEI (Health Effects Institute). 2009. Traffic‐Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Report #17. Available at: pubs.healtheffects.org/view.php?id=306.

Lucas K. 2004. Locating Transport as a Social Policy Problem. In: Lucas K Ed. Running on Empty: Transport, social exclusion, and environmental justice. Bristol, UK: Policy Press.

USEPA (US Environmental Protection Agency). 2001. Our Built and Natural Environments: A technical review of the interactions between land use, transportation, and environmental quality. Washington, DC: US Environmental Protection Agency.

3 USGAO (US General Accounting Office). 1999. Community Development: Extent of Federal Influence on Urban Sprawl is Unclear. GAO/RCED‐99‐87. Washington, DC: US General Accounting Office. Available at: www.gao.gov/archive/1999/rc99087.pdf.

4 Leyden KM. 2003. Social capital and the built environment: the importance of walkable neighborhoods. American Journal of Public Health 93:1546‐1551.

Transportation Research Board, Institute of Medicine of the National Academies: Committee on Physical Activity, Health, Transportation, and Land Use. 2005. Does the built environment influence physical activity? : examining the evidence. Special report 282. Washington, DC: Transportation Research Board.

5 Beck L, Dellinger A, O’Neil M. 2007. Motor vehicle crash injury rates by mode of travel, United States: using exposure‐based methods to quantify differences. American Journal of Epidemiology 166(2): 212‐218.

Pucher J, Dijkstra L. 2003. Promoting safe walking and cycling to improve public health: lessons from the Netherlands and Germany. American Journal of Public Health 93(9): 1509‐1516.

6 Petritsch T, Landis B, McLeod P, Huang H, Challa S, Guttenplan M. 2005. Level‐of‐service model for pedestrians at signalized intersections. Transportation Research Record 1939: 55–62.

Wier M, Weintraub J, Humphreys E, Seto E, Bhatia R. 2008. An area‐level model of vehicle‐pedestrian injury collisions with implications for land use and transportation planning. Accident Analysis & Prevention 41: 137‐145.

7 Zhou Y, Levy J. 2007. Factors influencing the spatial extent of mobile source air pollution impacts: a meta‐analysis. BMC Public Health 7(1): 89.

8 HEI (Health Effects Institute). 2009. Traffic‐Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Report #17. Available at: pubs.healtheffects.org/view.php?id=306.

Kim JJ, Huen K, Adams S, Smorodinsky S, Hoats A, Malig B, Lipsett M, Ostro B. 2008. Residential traffic and children's respiratory health. Environmental Health Perspectives 116(9): 1274‐1279.

McConnell R, Islam T, Shankardass K, Jerrett M, Lurmann F, Gilliland F, Gauderman J, Avol E, Künzli N, Yao L, Peters J, Berhane K. 2010. Childhood incident asthma and traffic‐related air pollution at home and school. Environmental Health Perspectives 118(7): 1021‐1026.

http://pubs.healtheffects.org/view.php?id=306




9 Seto EY, Holt A, Rivard T, Bhatia R. 2007. Spatial distribution of traffic induced noise exposures in a US city: an analytic tool for assessing the health impacts of urban planning decisions. International Journal of Health Geography 6: 24.

10 Miedema HM, Vos H. 1998. Exposure‐response relationships for transportation noise. Journal of the Acoustical Society of America 104: 3432‐3445.

Shield BM, Dockrell JE. 2003. The effects of noise on children at school: a review. Journal of Building Acoustics 10: 97‐106.

Stansfeld SA, Matheson MP. 2003. Noise pollution: non‐auditory effects on health. British Medical Bulletin 68: 243‐257.

Babisch W. 2008. Road traffic noise and cardiovascular risk. Noise and Health 10: 27‐33.

11 Gunier RB, Hertz A, Von Behren J, Reynolds P. 2003. Traffic density in California: socioeconomic and ethnic differences among potentially exposed children. Journal of Exposure Analysis and Environmental Epidemiology 13: 240‐246.

Green RS, Smorodinsky S, Kim JJ, McLaughlin R, Ostro B. 2004. Proximity of California public schools to busy roads. Environ Health Perspectives 112: 61‐66.

12 Cottrill CD, Thakuriah PV. 2010. Evaluating pedestrian crashes in areas with high low‐income or minority populations. Accident Analysis and Prevention 42(6): 1718‐1728.

LaScala E, Gerber D, Gruenwald PJ. 2000. Demographic and environmental correlates of pedestrian injury collisions: a spatial analysis. Accident Analysis and Prevention 32(5): 651-658. 13 NCHRP (National Cooperative Highway Research Program). 2011. Road Pricing: Public Perceptions and Program Development. NCHRP Report 686. Washington, DC: Transportation Research Board. Available at: ssti.us/wp/wp‐content/uploads/2011/08/nchrp_rpt_686.pdf.

14 USDOT (US Department of Transportation – Federal Highway Administration). 2011. Road Pricing Defined. Available at: www.fhwa.dot.gov/ipd/revenue/road_pricing/defined/index.htm.

15 City of Stockholm. 2006. Facts and Results from the Stockholm Trials, First version – June 2006. Available at: http://www.stockholmsforsoket.se/upload/The%20Stockholm%20Trial,%20facts%20and%20results_Expert%20Group%20Summary%20June%202006.pdf.

Transport for London. 2007. Central London Congestion Charging: Impacts Monitoring. Fifth Annual Report, July 2007. Available at: www.tfl.gov.uk/roadusers/congestioncharging/6722.aspx.

16 APHA (American Public Health Association). 2010. Health Impact Assessment: A Tool to Promote Health in Transportation Policy. Available at: www.apha.org/NR/rdonlyres/AB3486EF‐CA7F‐4094‐AE6E‐6AE87C6C26FB/0/HIATranFACTshtfinal.pdf.

17 Virginia Department of Transportation. 2010. Hot Lanes in the United States. Available at: virginiahotlanes.com/documents/Transurban‐HOT‐Lanes‐in‐the‐US‐Fact‐Sheetv2.pdf.

18 Congestion Pricing Vote Shot Down. 2008. Associated Press, Albany, New York. April 8, 2008.

19 Schaller B. 2010. New York City’s congestion pricing experience and implications for road pricing acceptance in the United States. Transport Policy 17: 266‐273. Available at: www.nyc.gov/html/dot/downloads/pdf/schaller_paper_2010trb.pdf.

20 Lovett K. 2011. Congestion pricing looks to hitch ride again with plan that could earn $1 billion for city. New York Daily News, September 18, 2011. Available at: http://www.nydailynews.com/ny_local/2011/09/18/2011‐09‐18_congestion_pricing_looks_to_hitch_ride_again.html#ixzz1Yb892IXl.

21 BMI (Battelle Memorial Institute). 2010. Synthesis of Congestion Pricing‐Related Environmental Impact Analyses: Final Report. Washington, DC: USDOT, FHWA. Available at: ops.fhwa.dot.gov/publications/fhwahop11008/fhwahop11008.pdf.

http://www.tfl.gov.uk/roadusers/congestioncharging/6722.aspx




Chin ATH. 1996. Containing air pollution and traffic congestion: transport policy and the environment in Singapore. Atmospheric Environment 30(5): 787‐801.

City of Stockholm. 2006. Facts and results from the Stockholm Trials, First version – June 2006. Available at: http://www.stockholmsforsoket.se/upload/The%20Stockholm%20Trial,%20facts%20and%20results_Expert%20Group%20Summary%20June%202006.pdf.

Transport for London. 2007. Central London Congestion Charging: Impacts Monitoring. Fifth Annual Report, July 2007. Available at: www.tfl.gov.uk/roadusers/congestioncharging/6722.aspx.

22 SFCTA. 2010. San Francisco Mobility Access and Pricing Study. San Francisco County Transportation Authority: San Francisco, CA. Available at: www.sfmobility.com.

23 SFCTA (San Francisco County Transportation Authority). 2010. San Francisco Mobility Access and Pricing Study. San Francisco County Transportation Authority, San Francisco, CA. Available at: www.sfmobility.com.

24 SFCTA (San Francisco County Transportation Authority). 2010. San Francisco Mobility Access and Pricing Study. San Francisco County Transportation Authority, San Francisco, CA. Available at: www.sfmobility.com.

25 SFCTA (San Francisco County Transportation Authority). 2010. Mobility, Access and Pricing Study Fact Sheet. San Francisco County Transportation Authority, San Francisco, CA. Available at: www.sfcta.org/images/stories/Planning/CongestionPricingFeasibilityStudy/PDFs/MAPS_fact_sheet_111010.pdf.

26 Kemm JR, Parry J, Palmer S. 2004. Health Impact Assessment: Concepts, Theory, Techniques, and Applications. USA: Oxford University Press.

27 North American HIA Practice Standards Working Group (Bhatia R, Branscomb J, Farhang L, Lee M, Orenstein M, Richardson M). 2010. Minimum Elements and Practice Standards for Health Impact Assessment, Version 2. Available at: www.sfphes.org/HIA_Tools/HIA_Practice_Standards.pdf.

28 Moscovich JL. 2008. Congestion pricing is a solution for San Francisco. San Francisco Chronicle (San Francisco, Ca) 12 December.

29 SFCTA (San Francisco County Transit Authority). 2010. San Francisco Mobility Access and Pricing Study. San Francisco, CA: San Francisco County Transportation Authority. Available at: www.sfmobility.com.

30 Congressional Budget Office. 2009. Using Pricing to Reduce Traffic Congestion. Congressional Budget Office, March 2009.

31 Environmental Defense Fund, National Association of City and County Health Officers, George Mason University. 2007. Are We Ready? Preparing for the Public Health Challenges of Climate Change. Available at: www.edf.org/documents/7846_AreWeReady_April2008.pdf.

Lee C, Abdel‐Aty M. 2005. Comprehensive analysis of vehicle‐pedestrian crashes at intersections in Florida. Accident Analysis and Prevention 37: 775‐786.

Miedema HME, Vos H. 1998. Exposure response for transportation. Journal of the Acoustical Society of America 104: 3432‐3445.

USEPA (U.S. Environmental Protection Agency). 2001. Vehicle travel: Recent trends and environmental impacts. Our Built and Natural Environments: A Technical Review of the Interactions Between Land Use, Transportation, and Environmental Quality. Chapter 4. Washington, DC: U.S. Environmental Protection Agency. Available at: www.epa.gov/smartgrowth/pdf/built_chapter3.pdf.

32 Appleyard D, Gerson MS, Lintell M. 1981. Livable streets. University of California Press, Berkeley, California.

33 Frank LD, Sallis JF, Conway TL, Chapman JE, Saelens BE, Bachman W. 2006. Many pathways from land use to health: Associations between neighborhood walkability and active transportation, body mass index, and air quality. Journal of the American Planning Association 72(1): 75‐87.

34 Besser LM, Dannenberg AL. 2005. Walking to public transit: Steps to help meeting physical activity recommendations. American Journal of Preventative Medicine 29(4): 273‐280.

http://www.edf.org/documents/7846_AreWeReady_April2008.pdf

http://www.epa.gov/smartgrowth/pdf/built_chapter3.pdf




35 Braveman P, Gruskin S. 2003. Defining equity in health. Journal of Epidemiology and Community Health 57: 254‐258.

36 Bullard RD. 2004. The Anatomy of Transportation Racism. In: Bullard RD, Johnson GS, Torres AO, eds. Highway Robbery: Transportation Racism & New Routes to Equity. Cambridge, MA: South End Press. 15‐31.

37 Bay Area Census, Association of Bay Area Governments and U.S. Census. 2000. Detailed Place‐to‐Place Commuting. Available at: www.bayareacensus.ca.gov/counties/SanFranciscoCounty.htm.

38 EDF (Environmental Defense Fund), National Association of City and County Health Officers, George Mason University. 2007. Are We Ready? Preparing for the Public Health Challenges of Climate Change. Available at: www.edf.org/documents/7846_AreWeReady_April2008.pdf.

39 SFCTA (San Francisco County Transit Authority). 2011. Modeling and Travel Forecasting. San Francisco, CA: SFCTA. Available at: www.sfcta.org/content/section/12/485/.

40MTC (Metropolitan Transportation Commission). 2004. San Francisco Bay Area Travel Survey 2000 Regional Travel Characteristics Report: Volume I. MTC: Oakland, California.

41 SFDPH (San Francisco County Transit Authority). 2011. Pedestrian Environmental Quality Index (PEQI). San Francisco, CA: SFDPH. Available at: www.sfphes.org/HIA_Tools_PEQI.htm.

42 Applied Geographic Solutions, Inc. Spring 2008 Update: Current Year Estimates. Methodology available at: www.appliedgeographic.com/library.html.

Applied Geographic Solutions, Inc. Spring 2009 Update: Future Year Estimates. Methodology available at: www.appliedgeographic.com/library.html.

43 CHP (California Highway Patrol), 2008. Accident Investigation Unit. Statewide Integrated Traffic Records System. Available at: www.chp.ca.gov/switrs/index.html.

44 Bhatia R. 2011. Health Impact Assessment: A guide for practice. In Press.

45 San Francisco Planning Department. 2009. Eastern Neighborhoods Area Plans. Available at: sf‐planning.org/index.aspx?page=1673

46 USEPA (U.S. Environmental Protection Agency). 2011. Clean Air Act. U.S. Environmental Protection Agency: Washington, DC. Available at: www.epa.gov/air/caa/.

47 Zhou Y, Levy JI. 2007. Factors influencing the spatial extent of mobile source air pollution impacts: a meta‐analysis. BMC Public Health 89: 1‐11.

Zhu Y, Hinds WC, Kim S, Shen S, Sioutas C. 2002. Study of ultra‐fine particles near a major highway with heavy‐duty diesel traffic. Atmospheric Environment 36: 4323‐4335.

48 Weijers EP, Khlystov AY, Kos GPA, Erisman JW. 2004. Variability of particulate matter concentrations along roads and motorways determined by a moving measurement unit. Atmospheric Environment 38: 2993‐3002.

49 Ross Z, Jerrett M, Ito K, Tempalski B, Thurston GD. 2007. A land use regression for predicting fine particulate matter concentrations in the New York City region. Atmospheric Environment 41: 2255‐2269.

Moore DK, Jerrett M, Mack WJ, Kunzli N. 2007. A land use regression model for predicting ambient fine particulate matter across Los Angeles, CA. Journal of Environmental Monitoring 9: 246‐52.

50 Delfino RJ. 2002. Epidemiologic evidence for asthma and exposure to air toxics: linkages between occupational, indoor, and community air pollution research. Environmental Health Perspectives 110(S4): 573‐589.

51 HEI (Health Effects Institute). 2009. Traffic‐Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Report #17. Available at: pubs.healtheffects.org/view.php?id=306.

52 Brunekreef B, Janssen NA, de Hartog J, Harssema H, Knape M, van Vliet P. 1997. Air pollution from truck traffic and lung function in children living near motorways. Epidemiology 1997(8): 298‐303.




53 Kim JJ, Smorodinsky S, Lipsett M, Singer BC, Hodgson AT, Ostro B. 2004. Traffic‐related air pollution and respiratory health: East Bay children’s respiratory health study. American Journal of Respiratory and Critical Care Medicine 170: 520‐526.

Lin S, Munsie JP, Hwang SA, Fitzgerald E, Cayo MR. 2002. Childhood asthma hospitalization and residential exposure to state route traffic. Environmental Research 88: 73‐81.

Venn AJ, Lewis SA, Cooper M, Hubbard R, Britton J. 2001. Living near a main road and the risk of wheezing illness in children. American Journal of Respiratory and Critical Care Medicine 164: 2177‐2180.

54 English P, Neutra R, Scalf R, Sullivan M, Waller L, Zhu L. 1999. Examining associations between childhood asthma and traffic flow using a Geographic Information System. Environmental Health Perspectives 107(9): 761‐767.

55 Gauderman WJ, Avol E, Lurmann F, Kuenzli N, Gilliland F, Peters J, McConnell R. 2005. Childhood asthma and exposure to traffic and nitrogen dioxide. Epidemiology 16(6): 737‐743.

Jerrett M, Shankardass K, Berhane K, Gauderman WJ, Künzli N, Avol E, Gilliland F, Lurmann F, Molitor JN, Molitor JT, Thomas DC, Peters J, McConnell R. 2008. Traffic‐related air pollution and asthma onset in children: a prospective cohort study with individual exposure measurement. Environmental Health Perspectives 116(10): 1433‐1438.

Kim JJ, Huen K, Adams S, Smorodinsky S, Hoats A, Malig B, Lipsett M, Ostro B. 2008. Residential traffic and children's respiratory health. Environmental Health Perspectives 116(9): 1274‐1279.

McConnell R, Berhane K, Yao L, Jerrett M, Lurmann F, Gilliland F, Kunzli N, Gauderman J, Avol E, Thomas D, Peter J. 2006. Traffic, susceptibility, and childhood asthma. Environmental Health Perspectives 114: 766‐772.

McConnell R, Islam T, Shankardass K, Jerrett M, Lurmann F, Gilliland F, Gauderman J, Avol E, Kuenzli N, Yao L, Peters J, Berhane K. 2010. Childhood incident asthma and traffic‐related air pollution at home and school. Environmental Health Perspectives 118(7): 1021‐1026.

56 Hoffmann B, Moebus S, Mohlenkamp S, Stang A, Lehmann N, Dragano N, Schmermund A, Memmesheimer M, Mann K, Erbel R, Jockel KH, Heinz Nixdorf Recall Study Investigative Group. 2007. Residential exposure to traffic is associated with coronary atherosclerosis. Circulation 116: 489‐496.

Hoffmann B, Moebus S, Stang A, Beck EM, Dragano N, Mohlenkamp S, Schmermund A, Memmesheimer M, Mann K, Erbel R, Jockel KH, Heinz Nixdorf Recall Study Investigative Group. 2006. Residence close to high traffic and prevalence of coronary heart disease. European Heart Journal 27: 2696‐2702.

57 USEPA (U.S. Environmental Protection Agency). 2011. EPA Clean Air Scientific Advisory Committee. Washington, DC: USEPA. Available at: yosemite.epa.gov/sab/sabpeople.nsf/WebCommittees/CASAC.

58 Moore DK, Jerrett M, Mack WJ, Künzli N. 2007. A land use regression model for predicting ambient fine particulate matter across Los Angeles, CA. Journal of Environmental Monitoring 9(3): 246‐52.

Ross Z, English PB, Scalf R, Gunier R, Smorodinsky S, Wall S, Jerrett M. 2006. Nitrogen dioxide prediction in Southern California using land use regression modeling: Potential for environmental health analyses. Journal of Exposure Science and Environmental Epidemiology 16(2): 106‐14.

Ross Z, Jerrett M, Itoc K, Tempalskid B, Thurston GD. 2007. A land use regression for predicting fine particulate matter concentrations in the New York City region. Atmospheric Environment 41(11): 2255‐2269

59 SFMTA (San Francisco Municipal Transit Authority). 2011. Muni Hybrid Buses. San Francisco, CA: SFMTA. Available at: www.sfmta.com/cms/mfleet/hybrids.htm.

60 Jerrett M, Burnett RT, Ma R, Pope CA 3rd, Krewski D, Newbold KB, Thurston G, Shi Y, Finkelstein N, Calle EE, Thun MJ. 2005. Spatial analysis of air pollution and mortality in Los Angeles. Epidemiology 16(6): 727‐736.

61 Johan de Hartog J, Boogaard H, Nijland H, Hoek G. 2010. Do the health benefits of cycling outweigh the risks?. Environmental Health Perspectives 118(8): 1109‐1116.




62 Dora C, Phillips M. 2000. Transport, environment and health. Copenhagen, Denmark: World Health Organization Regional Office for Europe.

Seto EY, Holt A, Rivard T, Bhatia R. 2007. Spatial distribution of traffic induced noise exposures in a US city: an analytic tool for assessing the health impacts of urban planning decisions. International Journal of Health Geography 6: 24.

63 Passchier‐Vermeer W, Passchier WF. 2000. Noise exposure and public health. Environmental Health Perspectives 108: 123‐131.

64 de Kluizenaar Y, Janssen SA, van Lenthe FJ, Miedema HM, Mackenbach JP. 2009. Long‐term road traffic noise exposure is associated with an increase in morning tiredness. Acoustical Society of America 126: 626‐633.

Passchier‐Vermeer W, Passchier WF. 2000. Noise Exposure and Public Health. Environmental Health Perspectives 108: 123‐131.

65 Shield BM, Dockrell JE. 2003. The effects of noise on children at school: a review. Journal of Building Acoustics 10: 97‐106.

66 Babisch W. 2008. Road traffic noise and cardiovascular risk. Noise and Health 10: 27‐33.

67 van Kempen EE, Kruize H, Boshuizen HC, Ameling CB, Staatsen BA, de Hollander AE. 2002. The association between noise exposure and blood pressure and ischemic heart disease: a meta‐analysis. Environmental Health Perspectives 110: 307‐317.


69 Bodin T, Albin M, Ardö J, Stroh E, Ostergren PO, Björk J. 2009. Road traffic noise and hypertension: Results from a cross‐sectional public health survey in southern Sweden. Environmental Health 8: 38.

70 WHO (World Health Organization). 1999. Guidelines for Community Noise. Available at: www.who.int/docstore/peh/noise/guidelines2.html.


Miedema H, Oudshoorn CGM. 2001. Annoyance from transportation noise: Relationships with exposure metrics DNL and DENL and their confidence intervals. Environmental Health Perspectives 109(4): 409‐416.

72 Physical Activity Guidelines Advisory Committee. 2008. Physical Activity Guidelines Advisory Committee Report, 2008. Washington, DC: U.S. Department of Health and Human Services.

73 WHO/Europe HEAT (Health Economic Assessment Tool). 2011. World Health Organization, Regional Office for Europe. Available at: http://www.heatwalkingcycling.org/index.php.

74 Active Living Research Program. 2009. Active Transportation: Making the Link from Transportation to Physical Activity and Obesity. Research Brief. Available at: www.activelivingresearch.org/files/ALR_Brief_ActiveTransportation.pdf

75 Transportation Research Board, Institute of Medicine of the National Academies: Committee on Physical Activity, Health, Transportation, and Land Use. 2005. Does the built environment influence physical activity?: examining the evidence. Special report 282. Washington, DC: Transportation Research Board.

76 Frank LD, Schmid TL, Sallis JF, Chapman J, Saelens BE. 2005. Linking objectively measured physical activity with objectively measured urban form. American Journal of Preventive Medicine 28(282): 117‐125.

Saelens BE, Handy SL. 2008. Built environment correlates of walking: A review. Medicine & Science in Sports & Exercise 40(7): S550‐S566.

Transportation Research Board, Institute of Medicine of the National Academies: Committee on Physical Activity, Health, Transportation, and Land Use. 2005. Does the built environment influence physical activity?: examining the evidence. Special report 282. Washington, DC: Transportation Research Board.




77 WHO/Europe Health Economic Assessment Tool (HEAT). 2011. World Health Organization, Regional Office for Europe. Available at: www.heatwalkingcycling.org/index.php.

78 City of Stockholm. 2006. Facts and results from the Stockholm Trials, First version – June 2006. Stockholm, Sweden: City of Stockholm. Available at: http://www.stockholmsforsoket.se/upload/The%20Stockholm%20Trial,%20facts%20and%20results_Expert%20Group%20Summary%20June%202006.pdf.

Transport for London. 2007. Central London Congestion Charging: Impacts Monitoring. Fifth Annual Report, July 2007. London, UK: Transport for London. Available at: www.tfl.gov.uk/roadusers/congestioncharging/6722.aspx.

79 Schoenborn CA, Stommel M. 2011. Adherence to the 2008 adult physical activity guidelines and mortality risk. American Journal of Preventative Medicine 40(5): 514‐521.

80 WHO/Europe Health Economic Assessment Tool (HEAT). 2011. World Health Organization, Regional Office for Europe. Available at: http://www.heatwalkingcycling.org/index.php.

81 Lee IM, Skerrett PJ. 2001. Physical activity and all‐cause mortality: what is the dose‐response relation? Medical Science Sports Exercise 33(6 Suppl):S459‐71; discussion S493‐4.

Schoenborn CA, Stommel M. 2011. Adherence to the 2008 adult physical activity guidelines and mortality risk. American Journal of Preventative Medicine 40(5): 514‐521.

82 NHTSA (National Highway Safesty Adminitaration). 2010. Traffic Safety Facts. US DOT – Washington, DC: NHTSA. Available at: www‐nrd.nhtsa.dot.gov/Pubs/811394.pdf.

83 Ewing R. 2006. Fatal and non‐fatal injuries. In: understanding the relationship between public health and the built environment: A report prepared for the LEED‐ND Core Committee. Available at: www.activeliving.org/files/LEED_ND_report.pdf.

Harwood D, Zegeer C, Lyon C, et al. 2008. NCHRP Web‐Only Document 129: Pedestrian Safety Prediction Methodology. National Cooperative Highway Research Program, Transportation Research Board, Washington, D.C. Available at: onlinepubs.trb.org/onlinepubs/nchrp/nchrp_w129p3.pdf.

NHTSA (National Highway Traffic Safety Administration). 1999. Literature Review on Vehicle Travel Speeds and Pedestrian Injuries. Washington, DC: DOT HS 809 021. Available at: http://www.nhtsa.gov/people/injury/research/pub/hs809012.html.

Retting RA, Ferguson SA, McCartt AT. 2003. A review of evidence‐based traffic engineering measures designed to reduce pedestrian‐motor vehicle crashes. American Journal of Public Health 93(9): 1456‐63.

84 Bhatia R, Wier M. 2011. “Safety In Numbers” Re‐examined: Can we make valid or practical inferences from available evidence? Accident Analysis & Prevention 43(1): 235‐240.

85 Brüde U, Larsson J. 1993. Models for predicting accidents at junctions where pedestrians and cyclists are involved. How well do they fit? Accident Analysis and Prevention 25(5): 499‐509.

Geyer J, Raford N, Ragland D, Pham T. 2006. The continuing debate about Safety in Numbers – Data from Oakland, CA. TRB Annual Meeting CD‐ROM.

Leden L. 2002. Pedestrian risk decrease with pedestrian flow. A case study based on data from signalized intersections in Hamilton, Ontario. Accident Analysis and Prevention 34: 457‐464.

Lee C, Abdel‐Aty M. 2005. Comprehensive analysis of vehicle‐pedestrian crashes at intersections in Florida. Accident Analysis and Prevention 37: 775‐786.

Loukaitou‐Sideris A, Ligget R, Sung HG. 2007. Death on the crosswalk: A study of pedestrian‐automobile collisions in Los Angeles. Journal of Planning Education and Research 26: 338‐351.

Lyon C, Persaud B. 2002. Pedestrian collision prediction models for urban intersections. Transportation Research Record 1818: 102‐107.

http://www.activeliving.org/files/LEED_ND_report.pdf

http://www.activeliving.org/files/LEED_ND_report.pdf




Wier M, Weintraub J, Humphreys EH, Seto E, Bhatia R. 2009. An area‐level model of vehicle‐pedestrian injury collisions with implications for land use and transportation planning. Accident Analysis and Prevention 41(1): 137‐145.

86 Cottrill CD, Thakuriah PV. 2010. Evaluating pedestrian crashes in areas with high low‐income or minority populations. Accident Analysis and Prevention 42(6): 1718‐1728.

87 Wier M, Weintraub J, Humphreys E, Seto E, Bhatia R. 2008. An area‐level model of vehicle‐pedestrian injury collisions with implications for land use and transportation planning. Accident Analysis and Prevention 41(1): 137‐145.

88 USDHHS (U.S. Department of Health and Human Services). 2000. Healthy People 2010: Understanding and Improving Health. 2nd ed. Washington, DC: U.S. Government Printing Office.

89 Wier M, Weintraub J, Humphreys E, Seto E, Bhatia R. 2008. An area‐level model of vehicle‐pedestrian injury collisions with implications for land use and transportation planning. Accident Analysis & Prevention 41(1): 137‐145.

90 Wier M, Weintraub J, Humphreys E, Seto E, Bhatia R. 2008. An area‐level model of vehicle‐pedestrian injury collisions with implications for land use and transportation planning. Accident Analysis & Prevention 41(1): 137‐145.

91 Elvik R. 2009. The non‐linearity of risk and the promotion of environmentally sustainable transport. Accident Analysis and Prevention 41(4): 849‐855.

92 Transport for London. 2007. Central London Congestion Charging: Impacts Monitoring. Fifth Annual Report, July 2007. London, UK: Transport for London. Available at: www.tfl.gov.uk/roadusers/congestioncharging/6722.aspx.

93 Sciortino S, Vassar M, Radetsky M, Knudson MM. 2005. San Francisco pedestrian injury surveillance: mapping, under‐reporting, and injury severity in police and hospital records. Accident Analysis and Prevention 37(6): 1102‐1113.

94 Richards DC. 2010. Relationship between Speed and Risk of Fatal Injury: Pedestrians and car occupants. Transportation research laboratory. Road safety web publication No. 16. London, UK: Department for Transport. Available at: www2.dft.gov.uk/pgr/roadsafety/research/rsrr/theme5/researchreport16/pdf/rswp116.pdf.

95 Beck LF, Dellinger AM, O’Neil ME. 2007. Motor vehicle crash injury rates by mode of travel, United States: Using exposure‐based methods to quantify differences. American Journal of Epidemiology 166(2): 212‐218.

96 Pucher J, Dijkstra L. 2003. Promoting safe walking and cycling to promote public health: lessons from the Netherlands and Germany. American Journal of Public Health 93(9): 1509–1516.

97 Elvik R. 2009. The non‐linearity of risk and the promotion of environmentally sustainable transport. Accident Analysis and Prevention 41(4): 849‐855.

98 SFMTA (San Francisco Municipal Transit Authority). 2010. City of San Francisco: 2010 Bicycle Count Report. San Francisco, California: SFMTA. Available at: 128.121.89.101/cms/bhome/documents/City_of_San_Francisco_2010_Bicycle_Count_Report_edit12082010.pdf.

99 Reynolds CC, Harris MA, Teschke K, Cripton PA, Winters M. 2009. The impact of transportation infrastructure on bicycling injuries and crashes: a review of the literature. Environmental Health 8: 47.

100 Lopez D, Sunjaya DB, Chan S, Dicker RA. Using trauma center data to identify missed bicycle injuries and their associated costs. In Press.

101 Braveman P. 2006. Health disparities and health equity: concepts and measurement. Annual Review of Public Health 27: 167‐194.

102 SFCTA (San Francisco County Transit Authority). 2010. Mobility, Access and Pricing Study Fact Sheet. San Francisco, CA: SFCTA. Available at: www.sfcta.org/images/stories/Planning/CongestionPricingFeasibilityStudy/PDFs/MAPS_fact_sheet_111010.pdf.




103 Brodin H, Hodge S. 2008. A guide to quantitative methods in Health Impact Assessment. Stockholm, Sweden: Swedish National Institute of Public Health. Available at: www.fhi.se/PageFiles/6057/R2008‐41‐Quantitative‐Methods‐in‐HIA.pdf.

USEPA (U.S. Environmental Protection Agency). 2000. Guidelines for Preparing Economic Analyses, Office of the Administrator. EPA‐240‐R‐00‐003. Washington, DC: U.S. EPA.

104 Moore JL. 1995. Cost‐benefit analysis: Issues in its use in regulation. Washington DC: Congressional Research Service.

Revesz RL. 1999. Environmental regulation, cost benefit analysis, and the discounting of human lives. Columbia Law Review 99: 941‐1017.

105 USEPA (U.S. Environmental Protection Agency). 2000. Guidelines for Preparing Economic Analyses, Office of the Administrator. EPA‐240‐R‐00‐003. U.S. EPA: Washington, DC.

106 Rice G, Hammitt J. 2005. Economic valuation of human benefits of controlling mercury emissions for U.S. coal fired power plants. Boston, MA: Harvard Center for Risk Analysis.

107 USDOT Memorandum on Treatment of the Economic Value of a Statistical Life in Departmental Analyses. Available at: http://ostpxweb.dot.gov/policy/reports/080205.htm.

108 Ackerman F, Heinzerling L. 2001. If it exists, it’s getting bigger: revising the value of a statistical life. Global Development and Environment Institute, Working Paper No. 01‐06. Medford, MA: Tufts University.

109 University of North Carolina Highway Safety Research Center. 2011. Pedestrian and Bicycle Information Center. Available at: www.bicyclinginfo.org.

110 VTPI (Victoria Transport Policy Institute). 2011. Transportation Demand Management Encyclopedia. Victoria Transport Institute, Victoria, BC. Available at: www.vtpi.org/tdm/.

111 San Francisco Planning Department. 2010. San Francisco Better Streets Plan. San Francisco, CA. Available at: www.sf‐planning.org/ftp/BetterStreets/proposals.htm#Final_Plan.

112 AASHTO. 2010. The Highway Safety Manual. Available at: www.highwaysafetymanual.org/Pages/default.aspx.

Ewing R. 1999. Traffic calming: State of the practice. Washington, DC: Institute of Transportation Engineers.

TCRP/NCHRP Report 112/562: Improving Pedestrian Safety at Unsignalized Crossings. 2006. Washington, DC: Transportation Research Board. Available at: onlinepubs.trb.org/onlinepubs/nchrp/nchrp_rpt_562.pdf.

113 AASHTO. 2010. The Highway Safety Manual. Available at: www.highwaysafetymanual.org/Pages/default.aspx.

TCRP/NCHRP Report 112/562: Improving Pedestrian Safety at Unsignalized Crossings. 2006. Transportation Research Board, Washington, DC. Available at: onlinepubs.trb.org/onlinepubs/nchrp/nchrp_rpt_562.pdf.

114 University of North Carolina Highway Safety Research Center. 2011. Pedestrian and Bicycle Information Center. Available at: www.bicyclinginfo.org.

http://www.vtpi.org/nmtguide.doc

Appendix A. Methods: Population Exposure Assignment for Air Quality and Noise Impacts Estimation Summary: The evaluation of exposure to air and noise pollution depends on the location of the source of emission as well as the location of the population. The following figure and summary provide an overview of the methods we used to estimate the residential population and residential population exposure for the air quality and noise HIA analyses.

AERMOD: U.S. Environmental Protection Agency recommended air quality modeling, further described in Appendix B.i TNM: Traffic Noise Model (TNM from the Federal Highway Administration, further described in Appendix C.ii

i Additional information available at: www.epa.gov/ttn/scram/dispersion_prefrec.htm#aermod. ii Additional information available at: www.fhwa.dot.gov/environment/noise/traffic_noise_model/tnm_v25/.

Appendix A. Methods: Population Exposure Assignment for Air Quality and Noise Impacts Estimation 1) Population Estimation Methods

Residential population at the Transportation Analysis Zone (TAZ) level (as detailed in Table 2 of the

report) was assigned to each residential land parcel in San Francisco based on an estimate of total

relative building floor area (area of parcel x average height of buildings) using the following formula:

Population of residential land parcel (A) = (B*C)/D

Where:

A = Population of residential land parcel

B = TAZ residential population

C = Building Floor Area estimate of residential land parcel

D = Total Building Floor of all residential land parcels in the specified TAZ

Residential land parcels were defined as parcels with a specified land use of residential or mixed land use (residential/commercial). For residential parcels, the entire parcel building floor area was used to calculate residential density. For mixed land use, we assume residential housing started on the second floor of the building. Land use information was obtained from the San Francisco Planning Department, based on data from the San Francisco Assessors Office and Department of Building Inspections. Additionally, residential building footprints were provided for the Presidio by Presidio GIS, Fort Mason by the Golden Gate National Recreation Area, and Treasure Island/Yerba Buena by Treasure Island Development Authority. For data not available on a TAZ level, proportions for race and age categories were derived from 2008

population and 2014 estimates (as specified in Table 2 of the report) and applied to the TAZ level data.

2) Air Quality and Noise Exposure Estimation As detailed in the Figure above, in a parallel process we estimated exposure levels to traffic‐related air

pollution (fine particulate matter, PM 2.5) and traffic‐related noise levels (as detailed in Appendix B and

C) under the three study scenarios (2005, 2015 BAU, 2015 RP) and assigned the maximum exposure

level to the population in the corresponding residential parcel. This use of maximum exposure tends to

overestimate health impacts, which is consistent with providing a health‐protective margin of safety in

impacts estimation. Additionally, the HIA describes relative estimated changes in health impacts across

the three scenarios – estimates in each of the scenarios based on the same assumptions. Appendices B

and C provide more information regarding the specific methods used to estimate health impacts.

Appendix B. Methods: Traffic‐Attributable Fine Particulate Matter (PM2.5) Exposure and Premature Mortality Summary: We estimated traffic‐attributable ambient fine particulate matter (PM2.5)

concentrations for the City and County of San Francisco for each of the three study scenarios

assigning exposure to the population at the level of the residential parcel. We then estimated

excess traffic‐attributable PM 2.5 mortality for each scenario utilizing residential populations

and exposure data, exposure‐response functions derived from well‐designed epidemiologic

studies, and current mortality incidence data for the City and County of San Francisco.

1. Estimation of Traffic‐Attributable Ambient PM2.5 We estimated traffic‐attributable PM2.5 concentrations for the three study scenarios (2005, 2015 BAU, 2015 RP) using the AERMOD air pollution dispersion model.a Model inputs included traffic volume data, speed, vehicle emissions, and meteorology. Traffic volume (annual average hourly) and speed data at the street segment level were estimated using the San Francisco County Transportation Authority’s SF‐CHAMP model (See Table 2 of the final report). We used the California Air Resources Board EMFAC 2007b model to estimate PM2.5 emissions in grams/mile. One year of hourly meteorological data (2008) was obtained from the Bay Area Air Quality Management District. The dispersion model estimated ambient concentrations city wide. 2. Health Impact Predictive Function A typical health impact predictive function has four components: 1) an effect estimate or exposure‐response function from an epidemiological or toxicological study; 2) an estimate for an exposure or exposure change; 3) a baseline incidence rate for the health effect; and 4) an estimate of the size of the population affected by exposure. Based on the population exposure estimated in each study scenario, we calculated excess traffic‐attributable PM2.5 mortality for each exposure category using the following equation.

Attributable Deaths = (e(‐β*δ PM2.5) – 1) * I O* Pexp Where: e(‐β*δ PM2.5) = an estimate of the relative risk of the incidence of mortality in a population exposed to a particular increment of PM2.5 above background levels β = coefficient of PM2.5 parameter in regression model δ PM2.5 = traffic attributable ambient concentration of PM2.5 I O = crude mortality incidence rate Pexp is the size of the population (> 30 years old) experiencing a change in exposure 3. Population Exposure to Traffic‐Attributable Ambient PM2.5 We assigned the modeled PM2.5 exposure to each parcel using the maximum modeled ambient exposure concentration at the parcel. We created discrete exposure categories for the range of traffic attributable PM2.5 concentration citywide. We estimated residential population data at the parcel level (estimated as detailed in Appendix A) to estimate the population exposed at each exposure category.

Appendix B. Methods: Traffic‐Attributable Fine Particulate Matter (PM2.5) Exposure and Premature Mortality 4. Baseline Disease and Mortality Incidence We utilized all‐cause mortality incidence data provided for the City and County of San Francisco by the California Department of Public Health. 5. Exposure Response Function The relationship between long term community‐level PM2.5 exposures over multiple years and community‐level annual mortality rates has been examined in several well‐designed, peer‐reviewed prospective cohort studies. These studies, conducted in the U.S. general population, report consistent statistically significant associations between PM2.5 and premature mortality. Three seminal studies based on two large general population cohorts (American Cancer Society and Harvard Six Cities) are summarized in the table below. The study by Jerrett et al (2005) re‐analyzed the American Cancer Society cohort using a different approach to exposure assignment and a subset of a cohort in Los Angeles, California. Key Cohort Studies of Long Term Exposure to PM2.5 and Mortality

Cohort/ Publication Population RR per 10 ug/m3 PM2.5 (95% Confidence Interval)

American Cancer Society Pope (2002)c

USA, 51 cities Adults, Age > 30 years

1.06 (1.02‐1.11)

Harvard Six Cities Laden (2006)d

USA, Multiple Cities General Population

1.16 (1.07, 1.26)

American Cancer Society Jerrett (2005)e

USA, Los Angeles Large Cohort (valid for comparison with large samples of U.S. Population)

1.17 (1.05 ‐1.30)

In developing exposure response functions for regulatory health impact assessment, the U.S. Environmental Protection Agency relies on the American Cancer Society (ACS) cohortf and the Harvard Six Cities cohort,g because of their geographic scope and their extensive reexamination.h Lower risk estimates from the ACS cohort may be due to higher population socio‐economic status or exposure misclassification from retrospective exposure assessments.i In the re‐analysis by Jerrett et al. (2005) of an ACS subpopulation in Los Angeles, more spatially refined intra‐regional exposure data may have reduced such exposure misclassification, potentially explaining a higher central relative risk estimate of 1.17 in the same population. Brunelle‐Yeung et al conclude that a RR=1.01 per 1μg/m3 PM2.5 (1% increase in mortality per 1μg/m3 of long‐term PM2.5) represents a central estimate of the exposure response function based upon the epidemiologic literature. j This exposure response function also represents the average of the individual estimates provided by 12 experts.k We utilized this exposure response function (RR=1.01 per 1μg/m3 PM2.5) in our health impact prediction equation.

Appendix B. Methods: Traffic‐Attributable Fine Particulate Matter (PM2.5) Exposure and Premature Mortality REFERENCES

a AERMOD Modeling System. 2011. Washington, DC: U.S. Environmental Protection Agency.

Available at: www.epa.gov/ttn/scram/dispersion_prefrec.htm#aermod. b EMission FACtors 2007 model (EMFAC 2007). 2010. California Air Resources Board. Available at: www.arb.ca.gov/msei/onroad/latest_version.htm. c Pope CA, Burnett RT, et al. 2002. Lung cancer, cardiopulmonary mortality, and long‐term exposure to fine particulate air pollution. Journal of the American Medical Association 287: 1132‐1141. d Jerrett M, Burnett RT, et al. 2005. Spatial analysis of air pollution and mortality in Los Angeles. Epidemiology 16: 727‐736. e Laden F, Schwartz J, et al. 2006. Reduction in fine particulate air pollution and mortality: Extended follow‐up of the Harvard Six Cities study. American Journal of Respiratory Critical Care Medicine 173: 667‐672. f Pope CA, Burnett RT, et al. 2002. Lung cancer, cardiopulmonary mortality, and long‐term exposure to fine particulate air pollution. Journal of the American Medical Association 287: 1132‐1141. g Jerrett M, Burnett RT, et al. 2005. Spatial analysis of air pollution and mortality in Los Angeles. Epidemiology 16: 727‐736. h Industrial Economics. 2010. Health and Welfare Benefits Analyses to Support the Second Section 812 Benefit‐Cost Analysis of the Clean Air Act. Washington, DC: U.S. Environmental Protection Agency. i Brunelle‐Yeung E, Masek T, Rojo JJ, Levy JI, Arunachalam S, Miller SM, Barrett SRH, Kuhn SR, Waitz IA. Methods for assessing the impact of aviation environmental policies on public health. Transport Policy In Press. j Brunelle‐Yeung E, Masek T, Rojo JJ, Levy JI, Arunachalam S, Miller SM, Barrett SRH, Kuhn SR, Waitz IA. Methods for assessing the impact of aviation environmental policies on public health. Transport Policy In Press. k Industrial Economics. 2006. Expanded Expert Judgment Assessment of the Concentration‐Response Relationship between PM2.5 Exposure and Mortality Final Report. Washington, DC: U.S. Environmental Protection Agency. Available at: www.epa.gov/ttn/ecas/regdata/ Uncertainty/pm_ee_report.pdf.

Appendix C. Methods: Road Traffic Noise and Noise‐related Annoyance and Myocardial Infarction Summary: We estimated sound levels for the City and County of San Francisco for the three scenarios

and assigned modeled ambient sound levels to residential parcels. We then estimated population

exposure levels to ambient sound based on the estimated number of residents of each parcel. We

estimated changes in noise‐related annoyance (defined below) using a well‐established exposure‐

response function based on data pooled from international studies. We estimated noise‐attributable

cases of myocardial infarction using an exposure response function derived from a meta‐analysis of

studies along with baseline myocardial infarction incidence rates for San Francisco, California.

1. Estimation of Ambient Sound Levels We estimated 24‐hour (Ldn) and daytime (Lday) ambient sound levels for the three study scenarios using the FHWA’s traffic noise model.a,b Model inputs included traffic volumes by time of day, vehicle type (car, truck), and traffic speeds (described in Table 2). Traffic also includes existing and future estimates of bus volumes. We assigned proportion of trucks that were heavy versus medium trucks based on neighborhood‐level estimates on arterial versus non‐arterial streets using aerial photography.c 2. Estimation of Population Exposure We used the residential population data at the parcel level (estimated as detailed in Appendix A) and assigned the modeled noise level exposure to each parcel using the maximum modeled noise exposure level within 25 meters of the parcel. 3. Health Impact Prediction Functions a) Noise Annoyance Noise annoyance is defined as “a feeling of resentment, displeasure, discomfort, dissatisfaction, or offense when noise interferes with someone's thoughts, feelings, or actual activities.”d Miedema (2001) pooled 45 international studies relating 24 hour noise levels and self‐rated noise annoyance and used these pooled data to develop exposure response functions for annoyance from road, railroad, and aviation noise.e The polynomial equation below estimates the proportion of the population reporting high levels of noise annoyance based on the 24‐hour sound level (Ldn) for road traffic noise. This function applies to a range of exposure from 42 dBA to 75 dBA.

42538.04210523.14210994.9% 2234 dndndn LLLHA

Using this function, we estimated the proportion of residents expected to be highly annoyed for 1dB sound level increments from 42dBA to 75dBA. We then multiplied these proportions by the number of people in each exposure category. The population exposed to noise levels above 75 dBA was assigned to the 75 dBA exposure category. b) Myocardial Infarction Babich (2008) pooled data from four well‐designed studies on noise and the incidence of myocardial infarction (MI) to derive an exposure‐response curve for MI incidence and daytime residential noise.f A polynomial function was fit to the pooled data for noise levels between 55‐ 80 dBA:

OR = 1.629657 – 0.000613*(Lday)2 + 0.000007357*(Lday)

3

Appendix C. Methods: Road Traffic Noise and Noise‐related Annoyance and Myocardial Infarction

We used this exposure‐response function to estimate the ORi for MI incidence for 1 dBA sound level increments from 55 dBA to 80 dBA. We then estimated the relative risk of MI incidence for each increment (RRi) using the following standard formula:

RR= OR/[1 + PMI *(OR ‐1)] Where PMI = 0.00182 (the baseline incidence of MI) We obtained baseline prevalence of myocardial infarction from a California Department of Public Health analysis of hospitalization discharge data reported to the Office of Statewide Health Planning and Development (OSHPD).g Data are collected from every licensed acute care hospital in California, excluding federal hospitals. We then estimated the fraction of MI cases attributable to noise exposure under each scenario using the standard formula for calculating attributable fractions for multiple category exposure.h,i

AFq = (Rq – 1) / Rq Where Rq= ∑i (qi RRi) and qi is the proportion of the population exposed at each exposure level. Finally, to estimate the annual number of traffic noise‐attributable MI cases, we multiplied the attributable fraction for each scenario by the product of the estimated population older than 30 years and the baseline MI incidence rate. REFERENCES

a Federal Highway Administration Traffic Noise Model, Version 2.5: Look‐Up Tables User’s Guide. 2004. USDOT, FHWA. Available at: http://www.fhwa.dot.gov/environment/noise/traffic_noise_model/tnm_v25_lookup/25lookup.pdf b Seto E, Holt A, Rivard T, Bhatia R. 2007. Spatial Distribution of Traffic Induced Noise Exposures in a U.S. City: an Analytic Tool for Assessing the Health Impacts of Urban Planning Decisions. International Journal of Health Geographics 6:24. c Holt A, Seto E, Rivard T, Gong P. 2009. Object‐based Detection and Classification of Vehicles from High‐resolution Aerial Photography. Photogrammetric Engineering & Remote Sensing 75:7. d Passchier‐Vermeer W, Passchier WF. 2000. Noise Exposure and Public Health. Environmental Health Perspectives 108:123‐131. e Miedema HM, Oudshoorn CG. 2001. Annoyance from transportation noise: relationships with exposure metrics DNL and DENL and their confidence intervals. Environ Health Perspectives 109(4):409‐16. f Babisch W. 2008. Road Traffic Noise and Cardiovascular Risk. Noise and Health 10:27‐33. g California Department of Public Health. Heart Attack (Myocardial Infarction) Data Query. Available at: http://www.ehib.org/page.jsp?page_key=126.

Appendix D. Methods: Active Transportation and Lives Saved from Walking and Cycling Walk and Bike Trips and Times by Age: We used existing and future conditions data from the

SFCTA model to assess walk and bike trips and trip times made by residents living in 12

geographic districts in San Francisco (the smallest reliable level of aggregation used to estimate

walk and bike trips in the SFCTA’s transportation analyses). We used data for existing and future

conditions for area‐level resident population and age as detailed in Table 2 of the final report.

As depicted in the map below, we collapsed the 12 districts into three districts, defined with by

their geographic location relative to the road pricing zone under study (depicted with a blue

boundary on the map). The three zones are “In the Zone,” turquoise, comprised of the districts

mostly in the cordon zone; “On the Fringe,” yellow/mustard, comprised of the districts just

outside of the cordon zone; and the “Outer Districts,” pink, representing those farther away

from the cordon zone.

Data were available on trip making and travel mode by district of residence but not by age in

San Francisco, so we estimated trip making and transportation mode by age based on the Bay

Area Travel Survey data (2000), assuming the proportion making trips and transportation mode

by age were similar across the Bay Area region.a

Lives Saved Due to Changes in Cycling and Walking: We used the WHO/Europe Health Economic Assessment Tool (HEAT) to estimate the number of lives saved to walking and cycling under the three study scenarios, which is readily available online. b

We used the trip and trip time estimates by age described above for residents in the three districts along with the annual death rate for San Francisco residents aged 25‐64 (267/100,000 residents) as inputs into the tool, which estimates annual lives saved from walking or cycling based on relative risk estimates applied to a population achieving a specified amount of annual physical activity from walking or cycling. For walking, the HEAT tool estimate is based on an epidemiologic meta‐analysis which found a relative risk estimate of 0.78 (95% confidence interval: 0.64‐0.98) for a walking exposure of 29 minutes seven days a week. For cycling, the HEAT tool is based on the relative risk data from three combined Copenhagen cohort studies which found a relative risk of all‐cause mortality of 0.72 among regular commuter cyclists aged


20‐60 years relative to the general population after controlling for socioeconomic variables (age, sex, smoking etc.) as well as for leisure time physical activity. REFERENCES

a Metropolitan Transportation Commission. 2004. San Francisco Bay Area Travel Survey 2000 Regional Travel Characteristics Report. Oakland, CA. b WHO/Europe Health Economic Assessment Tool (HEAT). 2011. World Health Organization, Regional Office for Europe. Available at: http://www.heatwalkingcycling.org/index.php.

http://www.heatwalkingcycling.org/index.php


Summary: We estimated census‐tract level changes in vehicle‐pedestrian injury collisions using

a multivariate, linear, area‐level regression model to predict the natural log of vehicle‐

pedestrian injury collisions in San Francisco census tracts.

1. 2005 Conditions: Vehicle‐Pedestrian Injury Collisions at the Census‐Tract Level

We estimated census‐tract level collisions for 2005 baseline conditions based on the average of

actual reported collisions for a five‐year period (2001‐2005) in an effort to reduce statistical

anomalies including regression to the mean. We used collision data from the Statewide

Integrated Traffic Records System (SWITRS) which contains data on reported vehicle collisions

on public roadways and is maintained by the California Highway Patrol. SWITRS vehicle‐

pedestrian injury collision data were imported into ArcGIS (version 9.2; ESRI Inc., Redlands,

California, USA) and geocoded to the intersection of the reported primary and secondary streets

(exact street address is not collected). We used a spatial join to assign vehicle‐pedestrian injury

collisions to one of the 176 census tracts in San Francisco. We excluded non‐injury collisions

which are reported as “Property Damage Only”. We included collisions resulting in pedestrian

injuries and/or fatalities, hereafter referred to as “vehicle‐pedestrian injury collisions”.

2. 2015 Conditions (Business As Usual (BAU), Road Pricing (RP)): Estimating Changes in

Vehicle‐Pedestrian Injury Collisions at the Census‐Tract Level

The San Francisco Department of Public Health developed a census‐tract level model of

pedestrian injury collision frequency as a function of aggregate traffic volume (log‐transformed),

street, land use and population characteristics. The model predicts over 70% of the variation in

vehicle‐pedestrian injury collision in San Francisco census tracts and is published in the peer‐

reviewed scientific journal Accident Analysis & Prevention. a

We used this multivariate, linear, area‐level (census tract) regression model developed in San

Francisco to estimate future changes in vehicle‐pedestrian injury collisions. The model predicts

census tract level vehicle‐pedestrian injury collision counts (log‐transformed) using the following

equation, where b0 is a constant and ∑biXi are significant area‐level model predictors including

traffic volume (log‐transformed), proportion arterial streets (without surface transit), proportion

of zoning that is neighborhood commercial or residential neighborhood commercial, number of

employees (log‐transformed), number of residents, land area, proportion of the population

living below poverty, and proportion of the population aged 65 and older.

ln(Ped Injury Collisions) = b0 + ∑biXi

Notably, several significant model predictors are potential proxies for pedestrian exposure:

employee and residential populations, the proportion of land area zoned neighborhood

commercial or residential‐neighborhood commercial (a pedestrian attractor), and the

proportion of people living in poverty (who are often more dependent on walking for

transportation).


We used the model to estimate 2005, 2015 BAU, and 2015 RP conditions, changing model

parameters based on estimated changes for those years in census‐tract level traffic volume,

number of residents and employees, proportion of the population aged over 65 and proportion

low income. Data for model covariates were obtained from the San Francisco County

Transportation Authority and census‐based estimates and planning projections from the City

and County of San Francisco and are further detailed in Table 2 of the final report. We used the

model predictions to calculate percent change in collisions from 2005 – 2015 BAU, 2005 – 2015

RP, and 2015 BAU – 2015 RP, and applied those percentages to the above 2005 conditions data

to estimate absolute changes in vehicle‐pedestrian injury collisions at the census tract level.

REFERENCES

a Wier M, Weintraub J, Humphreys EH, Seto E, Bhatia R. An area‐level model of vehicle‐pedestrian injury collisions with implications for land use and transportation planning. Accident Analysis & Prevention. 2009 Jan;41(1):137‐45.

Appendix F. Methods and Analyses: Pedestrian Environmental Quality Index (PEQI)

Summary: In response to community concerns regarding a lack of data on the pedestrian environment, the San Francisco Department of Public Health’s Program on Health, Equity and Sustainability developed the Pedestrian Environmental Quality Index (PEQI) to assess the quality of the physical pedestrian environment and inform pedestrian planning needs. The PEQI draws on published research and work from numerous cities to assess the physical environment, focusing on factors that impact whether people walk and the safety of the pedestrian environment in a neighborhood. The goal of the PEQI analysis is to examine walkability factors, to evaluate the existing quality of the pedestrian environment and to identify potential improvements during land use and transportation planning.

Methods The PEQI is an observational survey which quantifies street and intersection factors empirically known to affect people’s travel behaviors, and is organized into five categories: traffic, street design, land use, intersections, and safety. Within these categories are 30 indicators that reflect the quality of the built environment for pedestrians and comprise the survey used for data collection. SFDPH aggregates these indicators to create a weighted summary index, which can be reported as an overall index or deconstructed by pedestrian environmental category (see the following Table) or even by each indicator. Additional information regarding the PEQI, including a detailed methods report, and a data collection manual, the data collection form, and customize database, can be accessed online.i Table. PEQI Indicators by Pedestrian Environmental Category

Intersection Safety

Traffic Street Design Perceived Safety Land Use

Crosswalks

Ladder crosswalk

Countdown signal

Signal at intersection

Crossing speed

Crosswalk scramble

No turn on red

Traffic calming features

Additional signs for pedestrians

Number of vehicle lanes

Two‐way traffic

Vehicle speed

Traffic volume

Traffic calming features

Width of sidewalk

Sidewalk impediments

Large sidewalk obstructions

Presence of curb

Driveway cuts

Trees

Planters/gardens

Public seating

Presence of a buffer

Illegal graffiti

Litter

Lighting

Construction sites

Abandoned buildings

Public art/historic sites

Restaurant and retail use

The PEQI data are primarily collected with an observational surveyii based upon the visual assessment of street segments and intersections by a trained observer. A survey form is completed for each individual intersection and street segment (i.e., the segment of a street between two intersections). One indicator


– traffic volume – is not collected on the survey. For this variable SFDPH uses existing street segment traffic counts from the SFCTA’s SFCHAMP model. After data are collected for all desired streets and intersections, the PEQI data are entered into a database so that indicator responses can be converted into numeric values and then scored. Once data entry is complete, a series of programs applies weights to each indicator response, and calculates the PEQI summary scores for Street Segments and Intersections.

Indicator scores for each indicator category were created based on a survey of national experts, including city and transportation planners and consultants, and pedestrian advocates, regarding their importance to pedestrian environmental quality. To create the overall Street Segment, Intersection and other Domain PEQI scores (see Table, above), SFDPH aggregates the Indicator Response Category Weighted Scores and standardizes those PEQI summary scores so that the maximum score is “100” by multiplying that summed total by the corresponding Domain or Street Segment (Overall) weight. PEQI scores reflect the degree to which environmental factors supportive of walking and pedestrian safety have been incorporated into street segment and intersection design. The PEQI scores street segments and intersections separately, on a scale from 0 ‐100. SFDPH is currently using the following categories for scoring, a priori, with equal intervals of 20 points for all categories:

100‐81 = highest quality, many important pedestrian conditions present ‐ “Ideal”

80‐ 61 = high quality, some important pedestrian conditions present ‐ “Reasonable”

60‐ 41 = average quality, pedestrian conditions present but room for improvement ‐“Basic”

40‐ 21 = low quality, minimal pedestrian conditions – “Poor”

20 and below = poor quality, pedestrian conditions absent – “Not Suitable for Pedestrians” Detailed information regarding the indicator scores and this overall scoring approach is provided in the online report referenced above. A reliability study of the PEQI is currently underway. The next section includes a detailed summary of the PEQI analyses and findings for the Road Pricing HIA.


Analysis: Inside the Congestion Pricing Zone Figure A. Pedestrian Environmental Quality in the Congestion Pricing Zone

South of Market

In South of Market 198 intersections and 646 street segments were assessed using the PEQI (see Figure

A). The average PEQI intersection score was 28, which was the lowest score in the survey for area. Over

60 percent of the intersections lack basic pedestrian amenities and engineering countermeasures and

were ranked “not suitable for pedestrians” by the PEQI (red intersections). Many of these intersections

are junctions with short streets and alleys where there are no crosswalks, signals, or signs warning

pedestrians to look for cars; in some cases additional pedestrian infrastructure might be appropriate and

in other places not, based on factors including existing and projected pedestrian and traffic volumes.

The average PEQI street score for the streets was 47, also the lowest scoring neighborhood in this

survey. Twenty‐five percent of the streets were ranked “poor” (orange streets) or “not suitable for

pedestrians” (red streets). Many of the street segments that ranked “poor” are near the freeway, with


high traffic volumes and no traffic calming. Harrison is one street in particular that scored poorly

throughout South of Market.

Tenderloin

In Tenderloin 65 intersections and 216 street segments were assessed using the PEQI (see Figure A). The

average PEQI intersection score was 49. The majority of the intersection had “reasonable” pedestrian

conditions (light green) or “basic” (yellow, scores in the mid‐range). There were three intersections on

O’Farrell Street which ranked “not suitable for pedestrians” (red intersections). These intersections were

uncontrolled intersections, without crosswalk, traffic calming, pedestrian countermeasures and all

missing curb ramps. The average PEQI street score for the streets was 52, with the majority of streets

having “basic” conditions (yellow streets). There was one segment of Macallister Street which was

ranked “not suitable for pedestrians” (red street) due to missing sidewalks and no pedestrian amenities.

Civic Center

In Civic Center 58 intersections and 206 street segments were assessed using the PEQI (see Figure A).

The average PEQI score for the intersections was 42, with approximately one third of the intersections

lacking basic pedestrian amenities and engineering countermeasures and were ranked “not suitable for

pedestrians” by the PEQI (red intersections). The majority of these intersections were uncontrolled

intersections, without crosswalk, pedestrian counter measures and missing curb ramps. The

neighborhood streets have mostly “basic” (yellow streets) pedestrian environmental conditions with an

average score of 53. The streets that ranked poorly (orange streets) were all short streets and alleys.

Lower Nob Hill/Cathedral Hill/Lower Pacific Heights

In the Lower Nob Hill Area 65 intersections and 214 street segments were assessed using the PEQI (See

Figure A). The average PEQI score for the intersections was 39, with half of the intersections ranked “not

suitable for pedestrians” (red intersections) or “poor” (orange intersections) by the PEQI. These

intersections were all uncontrolled intersections without crosswalks, curb ramps and very little traffic

calming or pedestrian countermeasures. The streets have mostly “basic” pedestrian environmental

conditions (yellow, scores in the mid‐range) with an average score of 54. There were 12 streets that

ranked “poor” (orange streets), this included Austin Street and several other alleys and short streets.

Downtown/Union Square

In the Downtown/Union Square area 60 intersections and 208 street segments were assessed using the

PEQI (See Figure A). The average PEQI score for the intersections was 38, with half the intersections

ranked “not suitable for pedestrians” (red intersections) or “poor” (orange intersections) by the PEQI.

Many of these intersections are on Bush Street or around Union Square. The streets have mostly “basic”

(yellow streets) and “reasonable” (light green streets) pedestrian environmental conditions with an

average score of 55. There are two segments on Bush Street and several shorter streets that ranked

“poor” (orange street).


Financial District

In the Financial District 60 intersections and 150 street segments were assessed using the PEQI (see

Figure A). The average PEQI score for the intersections was 43, with most intersections having “basic

conditions” (yellow, scores in the mid‐range). The intersections that ranked “not suitable for

pedestrians” (red intersections) were mostly between 1st Street and New Montgomery Street and

several intersections on Bush Street, the majority of which were uncontrolled intersections. The

average PEQI score for the streets was 56, with approximately 75 percent of the streets having either

“basic” (yellow streets) or “reasonable” pedestrian conditions (light green streets). There were several

street segments around the Transbay Terminal that scored poorly, but can be partly contributed to the

construction.

Rincon Hill

In Rincon Hill 26 intersections and 92 street segments were assessed using the PEQI (see Figure A). The

average PEQI score for the intersections was 45, approximately a third of the intersections which ranked

“not suitable for pedestrians” (red intersections) or “poor” (orange intersections) by the PEQI were all

centered on alleys and freeway on and off ramps. The PEQI score for the streets was 47. While the

majority of the streets had “basic” (yellow streets) and “reasonable” (light green streets) pedestrian

quality, there were seven streets that did not have sidewalks and five streets that had sidewalks less

than five feet wide. The majority of streets ranked “poor” (orange streets) or “not suitable for

pedestrians” (red streets) were in close proximity to freeway on and off ramps.

South Beach

In South Beach 25 intersections and 96 street segments were assessed using the PEQI (see Figure A). The

average PEQI score for the intersections was 35, approximately half of the intersections ranked “not

suitable for pedestrians” (red intersections). These intersections tended to be uncontrolled intersections

connected to alleys. The PEQI score for the streets was 51; the majority of the streets that ranked

“poor” (orange streets) in this neighborhood were alleys. There were also several street segments on

Brannan Streets that scored “poor” (orange); all of these street segments had major sidewalk

impediments.

North Beach/Chinatown

In North Beach/Chinatown 57 intersections and 200 street segments were assessed using the PEQI (see

Figure A). The average PEQI score for the intersections was 33. As illustrated in the maps, almost half of

the intersections lack basic pedestrian amenities and engineering countermeasures and were ranked

“not suitable for pedestrians” by the PEQI (red intersections). Many of these intersections are junctions

with short streets and alleys, but there were also low scoring intersections on major roads, such as

Powell Street, Stockton Street and Grant Avenue. The PEQI score for the streets was 53. The

neighborhood streets have mostly “basic” (yellow, scores in the mid‐range) pedestrian environmental

conditions. A few have poor pedestrian conditions (orange), such as Broadway Street between Mason

Street and Powell Street, one segment of Powell Street and some of the alleys.


Analysis: Outside the Congestion Pricing Zone

Figure B. Pedestrian Environmental Quality along Geary Boulevard

Inner Richmond

In the Inner Richmond 78 intersections and 230 street segments were assessed using the PEQI (See

Figure B). Over half the intersections were ranked to have “reasonable” pedestrian conditions (light

green) or “basic” (yellow, scores in the mid‐range) pedestrian quality. Nine intersections were ranked

“not suitable for pedestrians” (red intersections). Many of these intersections were uncontrolled

intersection and lacked basic pedestrian amenities and engineering countermeasures .The

neighborhood streets have mostly “basic” (yellow streets) pedestrian environmental conditions with an

average score of 55. One street, Park Presidio Boulevard ranked “not suitable for pedestrians” (red

street). This street lacked sidewalks, was a multi‐lane road and had high traffic volumes.


Figure C. Pedestrian Environmental Quality in the Castro and the Mission

Mission Dolores/Castro/Duboce Triangle

In the Mission Dolores/Castro/Duboce Triangle area 71 intersections and 210 street segments were

assessed using the PEQI (See Figure C). The average PEQI score for the intersections was 40, although

over half of the intersections have “not suitable for pedestrians” (red intersections) or “poor” (orange

streets) environmental quality. Many of these intersections were uncontrolled intersections without

crosswalks. The PEQI score for the streets was 55. The neighborhood streets have mostly “basic”

(yellow, scores in the mid‐range) and “reasonable” pedestrian conditions (light green). A few have poor

pedestrian conditions (orange), such as Guerrero Street between 18th Street and 20th Street and some

of the alleys.

Mission

In The Mission 35 intersections and 124 street segments were assessed using the PEQI. The average

PEQI score for the intersections was 39. Over half of the intersections have “not suitable for


pedestrians” (red intersections) or “poor” (orange intersections) environmental quality. These

intersections are spread through the Mission on 23rd Street through 26th Street. The PEQI score for the

streets was 52, with most of the streets having “basic” (yellow, scores in the mid‐range) pedestrian

environmental conditions. The streets that ranked poor (orange streets) were mostly alley streets with

missing sidewalks.

REFERENCES

i SFDPH. The Pedestrian Environmental Quality Index. 2010. Available at: www.sfphes.org/HIA_Tools_PEQI.htm. ii SFDPH. Pedestrian Environmental Quality Index (PEQI) Survey Form. Available at: http://www.sfphes.org/publications/PEQI_Survey_Form.pdf.

http://www.sfphes.org/HIA_Tools_PEQI.htm


Summary: We estimated city‐wide changes in vehicle‐cyclist injury collisions using an accident

prediction equation derived from studies of vehicle‐cyclist collisions and published in the peer‐

reviewed journal Accident Analysis & Prevention (Elvik 2009). The equation requires only two

parameters – existing and future motor vehicle trips, and existing and future bicycle trips.

1. 2005 Conditions: Vehicle‐Cyclist Injury Collisions, Citywide and in the Cordon Zone

We estimated vehicle‐cyclist injury collisions for 2005 baseline conditions citywide using actual

reported collisions in collision data from the Statewide Integrated Traffic Records System

(SWITRS) which contains data on reported vehicle collisions on public roadways (CHP, 2008).

SWITRS vehicle‐cyclist injury collision data were imported into ArcGIS (version 9.2; ESRI Inc.,

Redlands, California, USA) and geocoded to the intersection of the reported primary and

secondary streets (exact street address is not collected). We excluded non‐injury collisions

which are reported as “Property Damage Only”. We included collisions resulting in cyclist

injuries and/or fatalities, hereafter referred to as “vehicle‐cyclist injury collisions”.

2. 2015 Conditions (Business As Usual (BAU), Road Pricing (RP)): Estimating Changes in

Vehicle‐Cyclist Injury Collisions, Citywide and in the Cordon Zone

We estimated future vehicle‐cyclist injury collisions using an accident prediction model equation

derived from studies of vehicle‐cyclist collisions and published in the peer‐reviewed journal

Accident Analysis & Prevention.a Model inputs are the number of motor vehicles and the

number of cyclists in the different study scenarios, as detailed in the following equation:

% Change in Vehicle‐Cyclist Collisions = (MVt2)0.7(CYCt2)

0.5

(MVt1)0.7(CYCt1)

0.5

Where:

MV = Number of motor vehicle trips

CYC = Number of Cyclist trips

t1 = Time period 1

t2 = Time period 2

We used this equation to estimate percent change in collisions from 2005 – 2015 BAU and from

2015 BAU – 2015 RP, based on projected changes in motor vehicle and cyclist trips using data

outputs from the San Francisco County Transportation Authority’s model. We then applied

those percentages to the above 2005 conditions data to estimate absolute changes in vehicle‐

cyclist injury collisions citywide and in the cordon zone.

We also conducted sensitivity analyses of the model estimates consistent with Elvik (2009),

varying the exponents as follows: MV = 0.8, CYC = 0.4; MV = 0.6, CYC = 0.7, and obtained very

similar estimates in the same direction of change.


REFERENCES

a Elvik R. The non‐linearity of risk and the promotion of environmentally sustainable transport. Accident Analysis & Prevention. 2009;41:849‐855.

Appendix H. Methods: Health Effects Characterization and Uncertainty Assessment Summary: Following is an excerpt from Health Impact Assessment: A Guide for Practiceadescribing the approach used to characterize estimated health effects and uncertainties for this HIA.

Characterizing expected effects on health

After the assessment team has gathered information and evidence and conducted analysis using data, they

will next need to synthesize their findings into a characterization of the expected health effects. There is no

gold standard for health effects characterizations in HIA and these characterizations are not testable or

falsifiable. The validity of characterizations rest on whether the assessment team’s judgments are plausible,

based on sound evidence, applies logical reasoning, and acknowledges data limitations and uncertainties.

(Veerman 2007; Petticrew 2007)

Four important and commonly described characteristics of health effects include likelihood, severity,

magnitude, and distribution. The likelihood of an effect represents the degree of certainty that it will

occur; likelihood is high when there is an established cause and effect relationship. The severity of a health

effect indicates its importance and intensity; for example, a disabling or life-threatening injury is more

severe than a self-limited infection. The magnitude represents how much a health outcome might change

as a result of a decision course of action. A magnitude may include the expected change in the population

frequency of symptoms, disease, illness, injury, disability, or reduced life-expectancy and is typically

estimated function of several factors including (1) the size of the population, (2) the baseline frequency of

disease, injury, illness, or mortality in the population, (3) the size of the change in the health risk or

resilience factor, and (4) the size or strength of association between an affected health risk factor and

health outcomes (e.g. the relative risk). The distribution of effects reflects whether they are shared fairly

among the affected populations. Each health effect characterization has a related assessment of the

uncertainty or limitations. (See step on uncertainty below)

In practice, these characteristics can be subject to varied interpretation both among members of the

assessment team and among stakeholders and decision-makers. Consequentially, discussion and debate on

the meaning and sufficiency of the characteristics and the evidence required to make a particular

characterization should be considered a necessary and useful part of a transparent HIA process.

Effect characterization in HIA may benefit from a common typology or nomenclature. The scheme

described in the table below provides one typology that can be used or adapted for diverse types of HIA.

In any such scheme, each health effect analyzed in HIA needs to be separately characterized. The HIA

should also explain where one or more health effect characteristics cannot be provided. Health effect characteristics and their interpretation

Likelihood How certain is it that the decision will effect health determinants or

outcomes irrespective of the frequency, severity, or magnitude Unlikely/ Implausible Logically implausible effect; substantial evidence against mechanism of effect

Possible Logically plausible effect with limited or uncertain supporting evidence Likely Logically plausible effect with substantial and consistent supporting evidence

and substantial uncertainties Very Likely / Certain Adequate evidence for a causal and generalizable effect

Insufficient Evidence / Not Evaluated

--

Appendix H. Methods: Health Effects Characterization and Uncertainty Assessment Severity How important is the effect with regards to human function, well being,

or longevity, considering the affected community’s current ability to manage the health effects

Low Acute, short term effects with limited and reversible effects on function, well-being, or livelihood that are tolerable or entirely manageable within the capacity of the community health system.

Medium Acute, chronic, or permanent effects that substantially affect function, well-being, or livelihood but are largely manageable within the capacity of the community health system; OR Acute, short-term effects on function, well-being, or livelihood that are not manageable within the capacity of the community health system

High Acute, chronic, or permanent effects that are potentially disabling or life-threatening, regardless of community health system manageability; OR Effects that impair the development of children or harm future generations.


--

Magnitude How much will health outcomes change as a result of the decision (i.e.,

what is the expected change in the population frequency of the symptoms, disease, illness, injury, disability, or mortality).

Limited A change of less than 1/10th of one percent in the population frequency of a health endpoint

Moderate A change of between 1/10th to one percent in the population frequency of a health endpoint

Substantial A change of greater than one percent in the population frequency of a health endpoint


--

Distribution Will the effects, whether adverse or beneficial, be distributed equitably

across populations. Will the decision reverse or undo baseline or historical inequities.

Disproportionate harms

The decision will result in disproportionate adverse effects to populations defined by demographics, culture, or geography

Disproportionate benefits

The decision will result in disproportionate beneficial effects to populations defined by demographics, culture, or geography

Restorative equity effects

The decision will reverse or undo existing or historical inequitable health-relevant conditions or health disparities


--

All health effect characterization should be supported with specific evidence.

Appendix H. Methods: Health Effects Characterization and Uncertainty Assessment Describing Sources of Uncertainty

HIA should always assess how gaps in evidence or assumptions may affect the

confidence in the characterization of health effects. Each fact used in

supporting an inference is a potential source of uncertainty. For example,

uncertainties in the baseline frequency of disease, the distribution of exposure,

or the relationship between exposure and disease all generate uncertainty in

health effect estimates derived from these parameters. Common simplifying

assumptions, for example, populations affected by a decision are similar to

study populations, may also be important sources of uncertainty.

A straightforward approach to characterizing the level confidence is to

identify each parameter that contributes to the effect characterization and

then to describe qualitatively the uncertainty in that parameter, explaining

how it may vary (be over- or under-estimated) and describing the influence of

such variation on the health effect conclusions. Table 6, below, illustrates the

conclusions of such an uncertainty analysis using the hypothetical example of

an HIA conducted on automated speed enforcement cameras discussed above.

When making quantitative estimates, sensitivity analysis (SA) can help

examine the relative importance of uncertain data inputs on predicted outcomes. While SA can employ

various techniques, a general approach is estimating effects while varying parameters that are considered to

be uncertain. Uncertainty in Health Effect Characterization of Automated Speed Enforcement

Health Effect Characteristic

Factors affecting certainty Confidence Level

Likelihood Established physical explanation of effects Consistent findings of empirical studies in diverse urban

contexts Effects in prospective interventions

High

Severity None High Magnitude Exposure Assessment

Distribution based on measured speed on a sample of 25, 30, and 35 mph city streets

Speeds aggregated into five mph increments Measures may under-represent high speed and volume

streets Distribution of measured speed may underestimate or

overestimate distribution of impact speed Baseline Disease Prevalence Pedestrian injury collisions are often under-reported Exposure Response Functions Speed-collision ERF not specific to pedestrian injury

collisions Speed-collision ERF not validated in study location Speed-severity ERF not validated in study location

Moderate

Distribution Not assessed N/A

In many ways, the task of

prediction in HIA is

analogous to the task of

diagnosis and prognosis

in clinical medical

practice where a

practitioner applies

training and experience, a

patient’s history, and

diagnostic tests. Prognosis

in medicine assumes

uncertainty and the

possibility of error, the

need for monitoring, and

f dj


REFERENCES

a Bhatia R. 2011. Health Impact Assessment: A Guide for Practice.

Appendix I. San Francisco Planning Neighborhood Boundaries and Names

Neighborhood boundaries defined by the San Francisco Planning Department.

Attachment A

Congestion pricing involves charging drivers a user fee to drive in specific, congested areas, and using the revenue generated to fund transportation improvements, such as better transit service, road improvements, and bicycle and pedestrian projects. The Mobility, Access, and Pricing Study (MAPS) examines the feasibility of a congestion pricing program in San Francisco, finding that it could be a highly effective way to manage our transportation system more efficiently and support the city’s future growth plans.

A solution for San Francisco’s congestion problemPeak-period traffic congestion in the city’s core areas negatively impacts our transportation system. During rush hour, congestion makes travel by both automobile and transit slow, with average auto speeds at 10 mph and average transit speeds at 8 mph in much of the downtown areas (see map, right). Congestion is already a problem today, and the city has ambitious growth plans for the future. Unless bold measures are taken, our streets will be unable to accommodate this growth without traffic coming to a standstill. A mobility and pricing program would send a signal to drivers to consider alternatives to driving in the peak and generate significant new revenue that would greatly improve transit service and the road network in San Francisco. Benefits from a potential congestion pricing program would include:

• 12% fewer peak period vehicle trips• 21% reduction in vehicle hours of delay• 5% reduction in greenhouse gases citywide• $60–80 million in annual net revenue for mobility improvements• 20–25% transit speed improvement• 12% reduction in pedestrian incidents

Given the high level of innovation associated with a congestion pricing program, a measured way to move forward is through a pilot project. A pilot would operate for 6–12 months and allow for real-time evaluation before the city considers anything permanent. Even then, several steps would be required before an implementation decision could be made.

Fact SheetLAST UPDATED

December 2010SAN FRANCISCO COUNTY TRANSPORTATION AUTHORITY

Mobility, Access, and Pricing Study

Revenues generated by congestion pricing would fund transportation improvements, such as better transit service, road improvements, and bicycle and pedestrian projects.

Contact Us!• www.sfmobility.com

• www.facebook.com/sfmobility

• www.twitter.com/sanfranciscota

• [email protected]

• 415.522.4819

• San Francisco County Transportation Authority 100 Van Ness Avenue, 26th Floor San Francisco, CA 94102 Attn: MAPS study

SOURCE: SFMTA APC data, 2007, and SFCTA LOS Monitoring, 2009.

Average travel speeds, PM peak hour

Congestion pricing and equityFor several reasons, we are confident that a congestion pricing program will not have an undue impact on low-income individuals:

• The Study recommends a 50% discount for low-income travelers;

• Only approximately 5% of peak-period travelers to downtown San Francisco are drivers with annual household incomes lower than $50,000;

• Support for studying a San Francisco congestion pricing program is highest among low-income Bay Area residents;

• Many of low-income travelers in the Bay Area are already on transit; congestion pricing will get transit moving faster, bringing all travelers faster, more reliable travel times.

Congestion pricing and the economyThe Study analyzed economic impacts of a program, finding that the program would have a neutral to positive impact for the following reasons:

• Congestion pricing would alleviate the severe economic costs caused by congestion in San Francisco today. More than half of the travel time for the average Bay Area auto trip can be attributed to congestion delay, translating to a cost of almost $7 per trip in travel time lost due to congestion, or over $2 billion for individuals and businesses in 2005. This is time and money that can be better used by Bay Area residents.

• More foot traffic is expected downtown, which could lead to more retail activity within the congestion pricing zone, as was the case when London and Stockholm implemented congestion pricing programs.

Effective ScenariosMAPS has identified three high-performing scenarios, each of which would be accompanied by substantial up-front transit improvements in corridors impacted by the fee. A future phase of the study will exclude additional analysis of the Southern Gateway scenario but will include designing an additional scenario that could achieve comparable benefits through a combination of downtown parking management policies and a modified cordon.

1. The AM/PM Northeast Cordon would collect $3/crossing from cars entering or leaving the zone in both AM and PM peak hours (capped at $6/day).

2. The PM-Outbound Northeast Cordon would collect $6/day from cars leaving the zone in the outbound direction during PM peak hours.

3. The Southern Gateway would collect $3/crossing from cars crossing the San Francisco-San Mateo county line in both directions in both AM and PM peak hours (capped at $6/day). It would also include increased parking rates downtown.

What does the public think? We have conducted four rounds of outreach over the course of the Study, reaching over 1,000 Bay Area residents through workshops, webinars, polls, focus groups, and other means.

Feedback from the most recent round of public meetings suggests support for continued evaluation of congestion pricing.

Permanent: I support implementing a permanent program.

Pilot: I prefer a pilot program approach.

Modified: I could support congestion pricing with modifications.

Not sure: I’m not sure yet/undecided/need more information.

Different solution: I prefer another solution.

2010 2011 2012 2013 2014 2015

Fall 2010–2011 Implement and

evaluate SFpark (SFMTA) and other

near-term projects.

2011–2013 Environmental analysis, system design; legislative authorization

2013–2014 Final Design and Procurement

2014–2015 Construction of system and capital improvements; additional transit services

2015 Potential implementation

IMPLEMENTATION DECISION

TimelineOn December 14, 2010, the Transportation Authority Board unanimously approved the MAPS Final Report and voted 8–3 in favor of pursuing additional study of the concept, excluding the Southern Gateway scenario.San Francisco is still in the early stages of exploring pricing and there are many steps before a decision will be made. tt

Winter 2010 Board adoption of MAPS final report

Excluded from further study

Attachment B

The San Francisco Model... in Fifteen MinutesWinter 2011

Usage Example: Golden Gate Bridge Destinations

Trips Diverted from Van Ness Avenue, with BRT Project

Usage Example: Trip Diversions from Van Ness Avenue, with Bus Rapid Transit (BRT) Project

Van Ness Diversions, 2010 Base to 2010 Project

Change in Volume at California St. 1700 100%

Traffic diverted onto:Parallel streets in corridor 800 47%Transit or suppressed trips 73 4%Parallel streets outside corridor 827 49%

Of traffic on streets outside corridor:About 1/2 are way beyond corridor (19th Ave, Kezar, etc.) 24%About 1/2 are evenly distributed across grid network 25%

Trips Diverted from Van Ness Avenue, with BRT Project (PM)

Thus about half of the traffic on Doyle westbound comes from

Van Ness Corridor. At Lombard, 60% of northbound

traffic on Van Ness is destined for Doyle Drive. At Grove, only

14% is destined for Doyle.

Green shows streets with less traffic.

Purple shows streets with more traffic.

The removal of one lane of through traffic from Van Ness is a 1/3 reduction in capacity on that roadway. The remaining two lanes are only slightly more congested – 71% of volume remains.

Due to the dense grid system and availability of wide parallel streets, the other streets within the corridor absorb almost 50% of the diverted traffic.

Other city streets absorb the remainder.

Regional Trips Local Trips Total

Divertible 1,823 (19%) 3,168 (33%) 4,991 (52%)Not Divertible 1,352 (14%) 3,239 (34%) 4,591 (48%)Total 3,175 (33%) 6.407 (67%) 9,582 (100%)

Doyle Drive Users (PM)

Less than 5% of vehicles on Doyle Drive (the access road connecting the Golden Gate

Bridge to the Marina District in San Francisco) are

traveling to or from the East Bay or South Bay.

The remainder are using Doyle to travel either across

town or over the Golden Gate Bridge. Less than 25% come

from Van Ness Avenue. Almost 30% use Franklin

Street.

How can models help us test policy questions? System efficiency... pricing strategies... demographic impacts... equity...

Models are just one tool in the planner’s kit that lets us compare alternatives, estimate impacts, assess ridership and benefit, and determine winners and losers.

No one tool can do everything – and no tool anywhere can replace good planning judgment. But, an effective model, used properly, is a sharp tool for decisionmaking.

Trip Based Models: Your Father’s Travel Forecasting Approach

Generation

What kinds of trips?How many?

Assignment

What specific route?

Mode Choice

What mode?Transit, Auto, Carpool

Nonmotorized?

Distribution

From where to where?

Tour Based Models:

Criticisms: No relationship between trips (in space, time, or mode!); nonmotorized modes usually missing; biased due to “zonal” aggregation

Tour based models predict activities, locations, and times.

A tour is an entire chain of trips, from a primary origin to all destinations, and then back to the primary origin.

Using chains means trips now have relationships:

(1) Primary Destination versus intermediate stops

(2) Consequences of mode availability: e.g. trip mode depends on the full tour mode:Leaving the car at home? Then no driving for any trip on that tour.

(3) Based on activity diaries for the household ... travel is implicit.

(4) More able to test key policy questionsno non-home-based trips e.g. system efficiency, pricing strategies,demographic impacts.

HOME

WORK

INTERMEDIATE STOP

EN ROUTE TO WORK

WORK-BASEDSUB-TOUR

DESTINATION

SECONDARYHOME-BASED TOUR

DESTINATION

1

7

6

5

4 3

2

= Tour

= Trip

Number indicates trip order

PRIMARY TOUR:Home-based Work

WORK-BASEDSUB-TOUR

SECONDARY HOME-BASED

TOUR HOME

WORK

INTERMEDIATE STOP

EN ROUTE TO WORK

WORK-BASEDSUB-TOUR

DESTINATION

SECONDARYHOME-BASED TOUR

DESTINATION

11

77

66

55

44 33

22

= Tour

= Trip


= Tour

= Trip


PRIMARY TOUR:Home-based Work

WORK-BASEDSUB-TOUR

SECONDARY HOME-BASED

TOUR

Microsimulation of Individual Behavior

While trip-based models aggregate travel into “zones” and then predict fractional probabilities for each zone and market segment, the San Francisco tour-based model simulates travel choices for every person in every household in the city.

The SF Model uses microsimulationof a “synthetic” population to do this. The dataset is “just like” the real people in San Francisco – based on recent U.S. Census data.

Trip Based Models:

Each market segment isa new set of trip tables.

Rows and columnsrepresent the “zones”that encompass the city.

Household Person Income Jobs Gender Age1 1 3 1 0 24

1 2 3 1 0 23

1 3 4 0 1 3

2 1 2 2 0 32

2 2 3 1 1 34

3 1 3 1 1 56

3 2 1 2 1 49

3 3 2 0 0 15

3 4 3 0 1 18

Tour Based Models:

Each market segment isa new column.

Every “synthetic” person in every household is present in the datafile and is modeled individually.

Model Development

Complete:

Value of time heterogeneity added for pricing sensitivityExpanded geographic bounds from San Francisco to whole Bay AreaUpdated accessibility variables

In development:

Bike Route Choice (mode choice sensitive to bike facilities, etc.)Land use growth allocation modelDynamic Traffic Assignment – Accounting for queues, bottlenecks, signal timingTravel Demand Management (TDM) Policies (i.e. fare subsidies)Discrete representation of Parking

Soon:

Refresh underlying data with Census/Survey data

Full-Day TourPattern Models

Time of DayModels

WorkplaceLocation Model

Vehicle AvailabilityModel

Nonwork TourDestination Choice

Models

Tour Mode ChoiceModels

Intermediate StopChoice Models

Trip Mode ChoiceModels

HighwayAssignment byTime Period (5)

TransitAssignment byTime Period (5)

Visitor Trip ModeChoice Model

Visitor Trip andDestination Choice

Model

LogsumVariables

AccessibilityMeasures

Zonal DataPopulationSynthesizer

Regional TripTables for NonSF

Trips

LogsumVariables

All Models

NetworkLevel of Service

All RemainingModels

Trip Diary Records

Person Records

Trip Tables

SF-CHAMP Model Process Diagram

Documents

Health Effects of Road Pricing In San Francisco, California · 2016-07-31 · Rajiv Bhatia Project Manager and Co‐Principal Investigator: Megan Wier Project Research Staff: Jennifer